IKS Health partnered with Wohlig Transformations to optimize the Stacks AI ML Engine — its AI-powered healthcare document-processing platform — across the OCR, Search, and LLM layers on Google Cloud, achieving $0.00557 per page (5.4× better than the SOW’s <3¢/page target) and a 15-percentage-point gain in multi-page document matching accuracy in a 3-week sprint.

Project Overview

As part of a wider cloud modernization initiative, IKS Health moved its Stacks AI ML Engine — the document-intelligence platform that processes patient records, dates of service, providers, medical images, and multi-page clinical reports — from a multi-vendor Azure AI Search + Azure OCR + OpenAI GPT-4o stack to a unified Google Cloud AI stack built on Vertex AI Search, Document AI, and Gemini 2.5 Flash. IKS Health had rebuilt the codebase on a five-service Cloud Run architecture before engaging Wohlig; our mandate was to tune the OCR, Search, and LLM layers and hand back a production-ready cutover plan. The work ran as a three-week optimization sprint (Apr 6 – Apr 24, 2026) plus cutover preparation into early May, with every artefact delivered to IKS Health’s ML and Server teams.

The Challenge

Multi-Vendor Complexity. AI workflows were split across Azure AI Search, Azure OCR, and OpenAI GPT-4o — three vendors, three SLAs, three cost models sitting in front of a single clinical workflow.

Healthcare Document Complexity. Production traffic includes multi-page clinical reports, MRI / X-ray imaging, mixed document types per batch, and multiple patients per submission — healthcare-grade accuracy is required on both single-page and multi-page workflows.

Large-File Processing Failures. High-resolution or high-page-count documents failed inside the initial pipeline even when on-disk file size was modest — a system-capacity issue, not a simple size threshold.

Cost-per-Page Target. The SOW set a hard ceiling of under 3 cents per page across the combined OCR + Search + LLM pipeline — end-to-end optimization, not single-component tuning.

Production-Grade Cutover. First-time go-live with no prior production to roll back to — every Go/No-Go gate had to be defensible.

Key Objectives

• Unified Cloud AI Stack: Migrate from Azure + OpenAI to Vertex AI Search + Document AI + Gemini 2.5 Flash without regressing on accuracy or relevance.

• Sub-3¢-per-Page Pipeline: Hit the SOW’s cost target across OCR + Search + LLM combined, on real benchmarked workloads.

• Multi-Page Accuracy Lift: Improve document grouping and field-level extraction on multi-page healthcare reports.

• Reusable AI Agent: Package the optimized engine as a self-contained, modular service IKS Health can drop into future projects.

• Production Cutover Plan: Go/No-Go gates, runbook, acceptance tests, and live-monitoring checks so first-time go-live behaves like a controlled deploy.

The Solution: Optimized Five-Service Cloud Run Pipeline

Five-Service Cloud Run Architecture. The engine runs as five independently-scaled Cloud Run services — chunk-coordinator → conversion-service (small + large) → ai-processing → status-notifier. CPU, memory, concurrency, timeout, and gunicorn workers are configured per service to match the workload.

The conv-large Split (Wohlig-introduced). We split conversion-service into two pools. Small jobs stay at 4 CPU / 16 GiB / concurrency 8. A new conversion-service-large runs at 8 CPU / 32 GiB, concurrency 1, timeout 3600s — a dedicated worker for high-resolution and high-page-count documents that isolates the long-tail without slowing the hot path.

OCRTEXT Pipeline Mode. We switched LLM_INPUT_TYPE from IMAGE to OCRTEXT — Gemini 2.5 Flash now consumes Document AI‘s clean OCR text instead of raw PNG buffers from every page. Smaller payloads, faster prompts, fewer hallucinations on tables and dates — the single change that lifted both accuracy and cost together.

Gemini 2.5 Flash (fine-tuned). Replaces OpenAI GPT-4o at the LLM stage. Cross-model benchmarking across GPT-4o, Gemini Pro, and Gemini 2.5 Flash drove the selection on unit cost, latency, and field-level accuracy; Gemini Pro stays available as a fallback for harder document classes.

Technology Stack. Vertex AI Search, Document AI, Gemini 2.5 Flash, Cloud Run, GKE, BigQuery, Cloud Storage, Cloud Build, Terraform, Cloud Monitoring, Cloud IAM, Secret Manager.

Key Benefits & Results

• Previous: $0.00689 per page on the Azure + OpenAI baseline. Our Solution: OCRTEXT mode + fine-tuned Gemini 2.5 Flash + per-service Cloud Run sizing. Result: $0.00557 per page — 5.4× better than the SOW’s <3¢ target; total run cost $7.736 → $6.223 on the same 1,118-page benchmark (−19.56%).

• Previous: Multi-page document matching at 65%. Our Solution: conv-large split + chunking + OCRTEXT pipeline. Result: 80% multi-page matching (+15 percentage points).

• Previous: Files Fully Matched at 67.60%. Our Solution: Optimized end-to-end pipeline. Result: 78.10% (+10.50 pp).

• Previous: Field-level accuracy on the Azure baseline. Our Solution: Field-specific prompts + OCRTEXT mode. Result: PatientDOB 82.09% → 89.94% (+7.85 pp), PatientName 95.06% → 97.69%, DateOfService 66.10% → 70.00%, Provider 55.75% → 58.78%.

• Previous: Large-file processing failures on high-resolution / high-page-count documents. Our Solution: New conversion-service-large (8 CPU / 32 GiB / concurrency 1 / 3600s). Result: Mitigated and tracked through cutover gate G5.

• Previous: No production cutover discipline. Our Solution: 11 Go/No-Go gates + 14-step runbook + 6 acceptance tests + 8 live-monitoring checks. Result: Production-grade first-time go-live readiness, handed to the ML and Server teams.

Technical Innovation

OCRTEXT Pipeline Mode. Switching Gemini’s input from raw PNG buffers to Document AI OCR text simultaneously lifted accuracy and dropped cost — a single change with two-axis impact.

conv-large Service Split. A dedicated Cloud Run service for high-resolution and high-page-count documents (8 CPU / 32 GiB / concurrency 1), without sacrificing throughput on the small / fast documents that route to conv-small.

Tuned Per-Service Sizing. Each of the five Cloud Run services is configured independently — CPU, memory, concurrency, timeout, gunicorn workers and threads — for the workload it actually handles.

Production Cutover Discipline. 11 Go/No-Go gates, a 14-step runbook, 6 acceptance tests, and 8 live-monitoring checks. Open production-readiness items (large-file, high-page-count, bulk-load) are tracked transparently as gates G5–G7.

Reusable AI Agent Packaging. The optimized stack ships as a self-contained, modular agent that IKS Health can drop into future document-intelligence workflows without re-architecting the OCR / Search / LLM layer.

Wohlig’s Approach

1. Discovery & migration assessment — audit existing Azure AI Search, Azure OCR, and OpenAI GPT-4o integrations; document baselines, token usage, and the per-page cost scoreboard.

2. Vertex AI Search optimization — tune data stores, schemas, and ingestion pipelines; validate content and embedding parity with the Azure baseline.

3. Document AI OCR integration — configure processors for IKS Health’s document types; switch the pipeline to OCRTEXT mode.

4. Gemini LLM optimization & prompt re-engineering — re-engineer prompts for Gemini’s instruction format and 1M-token context window; fine-tune Gemini 2.5 Flash for stable field-level extraction.

5. Code refactoring & reusable agent packaging — factor the orchestration into a modular, agent-based service IKS Health can reuse across future projects.

6. Cross-model benchmarking, system testing, cutover plan — GPT-4o vs Gemini Pro vs Gemini Flash on the same workload; an 11-scenario test pass (single-page, multi-page grouping, improper sequence, mixed document types, missing fields, fuzzy matching, exact-match DOB / DOS, case insensitivity, MRI / X-ray imaging, low-quality OCR, multi-patient batches); production cutover plan with 11 Go/No-Go gates, 14-step runbook, 6 acceptance tests, and 8 live-monitoring checks.

About IKS Health

IKS Health is a leading US-focused healthcare solutions company providing revenue cycle management, clinical documentation improvement, and IT services to hospitals and physician groups. By combining clinical expertise with intelligent automation, IKS Health reduces administrative burdens so clinicians can focus on delivering quality patient care.

About Wohlig Transformations Pvt. Ltd.

Founded in 2015, Wohlig Transformations specialises in GenAI and DevOps, with 160+ professionals across India and the UK.

Detailed Case Study : https://youtu.be/CV7GvA5VonU?si=k7F0zTs74Ts2bgVz

Mahindra & Mahindra partnered with Wohlig Transformations to build an AI-powered Vehicle Development, Benchmarking, and Product Planning Platform on Google Agent Development Kit (ADK) + Gemini 3 Flash, evolving from a 90-spec MVP into a production platform analysing 207 specs per car across three integrated functions.

Project Overview

Mahindra & Mahindra‘s R&D, product-planning, and competitive-intelligence teams partnered with Wohlig Transformations to put AI at the centre of how vehicle decisions get made. The engagement began in February 2026 with a focused 3-week SOW: an ADK-powered automotive benchmarking agent covering 90 car specs, car-only competitor analysis, and integration with Mahindra’s existing ChatAI on Google Cloud. The MVP proved the approach quickly — and on the strength of that impact, Mahindra expanded the scope substantially. The production platform now covers 207 specs per car across three integrated functions — Vehicle Development, Benchmarking, and Product Planning — with a custom React.js UI/UX, ADK exposed as the REST API backend, user RBAC (admin / analyst / viewer), custom views per spec or spec group, chat history across sessions, and multi-file upload for document intelligence. A multi-source data layer — Google Custom Search API, web scraping, YouTube Data API v3, and PDF document intelligence — feeds Gemini 3 Flash agents, all deployed on Google Cloud.

The Challenge

Single-Function Bottleneck: The original ChatAI integration only supported competitor benchmarking — but Mahindra’s R&D teams needed an AI workstation spanning vehicle development, benchmarking, and product planning, not a point tool.

Limited Spec Coverage: 90 car specs was the right scope to validate an MVP, but production decisions across SUVs, pickups, commercial vehicles, and tractors demand far richer analytical depth.

Multi-Source Data Sprawl: Vehicle insights live across official automotive sites, expert review portals, YouTube channels, and OEM PDF brochures — no single source covers it all, and stitching them together manually doesn’t scale.

RBAC + Custom Views: Admins, analysts, and viewers each need different surfaces — and analysts often need bespoke spec groupings that a default schema can’t anticipate.

Conversational Continuity: One-off queries weren’t enough; teams needed chat history that preserved investigation context across multi-day workflows.

Key Objectives

• Expand Spec Coverage: Grow from 90 to 207 specs per car to support deep R&D decisions.

• Three Integrated Functions: Unify Vehicle Development + Benchmarking + Product Planning in one platform.

• Custom React UI/UX: Replace the existing-ChatAI integration with a purpose-built frontend driven by ADK as a REST API.

• RBAC + Custom Views: Admin / analyst / viewer tiers plus per-user custom views of individual specs or spec groups.

• Multi-Source Data Integration: Combine web scraping, Google Custom Search API, YouTube Data API v3, and PDF document intelligence with RAG.

• Production Security: Cloud IAM, Secret Manager, API Gateway, end-to-end encryption, and VPC isolation.

The Solution: AI-Powered Vehicle Development + Benchmarking + Product Planning Platform

V1 — The 3-Week MVP (Feb–Mar 2026): An ADK-powered benchmarking agent covering 90 car specs and car-only competitor analysis, integrated with Mahindra’s existing ChatAI. A Python backend on Google Cloud, with RBAC, interactive dashboards, and exportable reports (PDF / Excel / PPT) — delivered in three weeks.

V2 — The Expanded Production Platform: 207 specs per car across three integrated functions — Vehicle Development, Benchmarking, and Product Planning. A custom React.js UI/UX with ADK as the REST API backend, RBAC plus custom views per spec or spec group, chat history across sessions, and multi-file upload with document intelligence (OCR + RAG).

Multi-Source Data Layer: Google Custom Search API (domain-whitelisted to curated automotive sources) + web-scraping pipelines + YouTube Data API v3 (video reviews, expert opinions, sentiment) + PDF brochure ingestion → OCR → vector embeddings → RAG corpus.

AI Core: Vertex AI Gemini 3 Flash agents orchestrated by ADK in a multi-agent pipeline — competitor data extraction → spec normalisation → RAG → comparative insights → report drafting.

Technology Stack: Vertex AI, Gemini 3 Flash, Agent Development Kit (ADK), React.js, Python, FastAPI, Google Custom Search API, YouTube Data API v3, Cloud Run, Cloud Storage, BigQuery, Cloud IAM, Secret Manager, API Gateway, and Cloud Operations Suite.

Key Benefits & Results

• Previous: 90 specs, car-only benchmarking. Our Solution: 207-spec, 3-function platform (Vehicle Development + Benchmarking + Product Planning). Result: 2.3× analytical depth and coverage of the full R&D decision lifecycle.

• Previous: Integration with existing ChatAI (limited UX flexibility). Our Solution: Custom React.js UI/UX with ADK as REST API. Result: Frontend and backend evolve independently — a future-proofed architecture.

• Previous: RBAC tiers only. Our Solution: RBAC + per-user custom views per spec or spec group. Result: Analysts and product planners get bespoke surfaces without admin intervention.

• Previous: Stateless queries. Our Solution: Chat history across sessions. Result: Investigation context preserved across multi-day workflows.

• Previous: Single PDF upload. Our Solution: Multi-file upload. Result: Document intelligence across multiple brochures and reports in one session.

• Previous: Single-source competitor data. Our Solution: Multi-source integration (Custom Search + Web Scraping + YouTube + Document RAG). Result: Richer, cross-validated insights for product decisions.

• Previous: Manual report drafting. Our Solution: AI-generated comparative reports. Result: Exportable PDF / Excel / PPT outputs ready for stakeholder review.

Technical Innovation

ADK as REST API + React Frontend: A modular architecture where Google’s Agent Development Kit is exposed as a backend microservice and the frontend is a custom React.js application. Frontend and backend evolve independently — Mahindra can change UX patterns without touching agent logic, and Wohlig can swap models or pipelines without touching the UI.

207-Spec Schema with Custom Views: A full automotive specification schema (engine, safety, comfort, infotainment, dimensions, performance, and more) at 207 fields per car, with user-defined custom views that let analysts pivot on any combination of specs or spec groups.

Multi-Source AI Pipeline: Google Custom Search API (domain-whitelisted to curated automotive sources) + web scraping + YouTube Data API v3 + PDF document intelligence with OCR + RAG — four orthogonal data streams unified by ADK orchestration and Gemini 3 Flash insight generation.

RBAC + Chat History + Multi-File Upload: Production-grade experience features — admin / analyst / viewer tiers, persistent chat history, and multi-document upload with RAG — layered on top of the agent, and uncommon in MVP-stage automotive AI tools.

Continuous Scope Expansion: A 3-week SOW MVP that grew into a sustained production engagement. The V1 → V2 jump — 90 → 207 specs, single → three functions, integration → custom UI — demonstrates Wohlig’s ability to evolve a product alongside client adoption.

Wohlig’s Approach

1. Architecture & setup — GCP project setup, access finalization, and platform integration scoping.

2. AI & data pipeline development — Web scraping + Google Custom Search API + YouTube Data API v3 integrations, Gemini 3 Flash workflows, and ADK orchestration.

3. Visualization & reporting — Interactive dashboards, report templates, and multi-format export (PDF / Excel / PPT).

4. V1 deployment & UAT — Cloud Run deployment, Cloud IAM + RBAC, monitoring, and Mahindra UAT.

5. V2 scope expansion — 207-spec schema; three functions (Vehicle Development + Benchmarking + Product Planning); React.js UI/UX rebuild; ADK as REST API; custom views; chat history; multi-file upload.

6. Continuous iteration — Documentation, training, knowledge transfer, and ongoing optimization with Mahindra’s R&D and product-planning teams.

About Mahindra & Mahindra

Mahindra & Mahindra Ltd. is one of India’s largest multinational automotive corporations, headquartered in Mumbai. Part of the globally diversified Mahindra Group, its portfolio spans SUVs, pickup trucks, commercial vehicles, tractors, and emerging mobility solutions across India, South Africa, Australia, and Latin America. Mahindra’s R&D and product-planning functions are leading AI adoption to accelerate vehicle development and competitive intelligence.

About Wohlig Transformations Pvt. Ltd.

Founded in 2015, Wohlig Transformations specialises in GenAI and DevOps, with 160+ professionals across India and the UK.

Detailed Case Study : https://youtu.be/aWMv4Ff_r2g

Project Overview

Dr. Reddy’s Laboratories (DRL) partnered with Wohlig Transformations in a two-phase engagement — starting with a 1-week Google Cloud GenAI workshop and continuing into an ongoing engineering engagement building Project Cognito, DRL’s AI-powered drug prioritization platform. The first production pillar shipped is the IP (Intellectual Property) Pillar — a multi-model patent-landscape analysis pipeline (Gemini extracts, Claude judges) running across four analytical dimensions.

DRL’s R&D, Manufacturing, Quality, and Biologics functions — led by Nishit Mittal as Data Science Lead — engaged Wohlig to accelerate GenAI adoption across drug-development decision-making. The work followed a deliberate Workshop → Production arc: a 1-week capability demonstration first, then a continuous production-engineering relationship. In that second phase, Wohlig is building Project Cognito, DRL’s umbrella platform for drug prioritization and research automation, delivered as discrete production pillars. The first to ship is the IP Pillar, a multi-model pipeline that pairs Gemini for extraction with Claude as a judge across four analytical dimensions, runs on 10 parallel Cloud Run instances, retrieves from a ChromaDB vector store, and refreshes automatically on a bi-weekly schedule.

The Challenge

Capability Demonstration

Before committing to a long-term AI engineering engagement, DRL’s R&D leadership needed to see Wohlig build production-grade patterns end-to-end on real pharma use cases — not slideware.

Patent Landscape Complexity

A drug’s IP exposure spans composition-of-matter, formulation, device, and process patents — each with different inclusion logic, jurisdictional nuances, and litigation history.

Multi-Source Data Sprawl

Patent data lives across Espacenet, Google Patents, and the Indian Patent Database (IPD); clinical evidence spans 6+ international registries. Each source has its own schema, latency, and gaps.

LLM Fragility

Single-model patent analysis hallucinates classifications, misses contextual Tier-3 matches, and produces malformed JSON — none of which is acceptable in a system informing real drug-investment decisions.

Production Scale

A bi-weekly refresh across hundreds of drugs requires parallel compute, retry strategies, incremental processing, and cost discipline — far beyond what a notebook prototype provides.

Key Objectives

• Workshop-First Demonstration: Build all three workshop modules end-to-end on Google Cloud (ADK + Vertex AI + Document AI + Vector Search + Cloud Run).

• Multi-Model Verification: Use Gemini for extraction and Claude as judge to catch hallucinations on every field.

• Tiered Patent Inclusion: Codify a Tier 1 / Tier 2 / Tier 3 taxonomy that surfaces every relevant patent, including non-obvious contextual matches.

• Multi-Source Coverage: Index every relevant patent and clinical-trial source (Espacenet, Google Patents, IPD, ClinicalTrials.gov, PubMed, ChiCTR, EU CTR, CTRI India, JRCT Japan).

• Production Compute: Parallel Cloud Run pipelines, Cloud Scheduler refresh, BigQuery storage, and an AlloyDB migration path.

• Continuous Optimization: Cost monitoring, per-dimension evaluation metrics, and knowledge transfer to DRL.

The Solution: Two-Phase GenAI Engagement

Phase 1: 1-Week Workshop

A Google Cloud GenAI workshop delivered three hands-on modules end-to-end.

Module 1 was a multi-agent Intelligent Chatbot built on four ADK agents (Structured Data, Unstructured Data, Web Search, and a Response Aggregator) with custom RAG on Vertex AI Vector Search.

Module 2 was a Document Intelligence pipeline for FDA Complete Response Letter (CRL) analysis using Document AI plus four specialized agents (Checklist, Summary, Metadata, Cross-Reference).

Module 3 was an MCP (Model Context Protocol) server giving a natural-language interface to BigQuery and Cloud SQL, containerised on Cloud Run.

Phase 2: Project Cognito

In the ongoing engagement, Wohlig built Project Cognito’s IP Pillar end-to-end, scaling the workshop’s proven patterns into a production system.

IP Pillar Architecture

Gemini extracts, Claude judges; 10 parallel Cloud Run instances per run; a ChromaDB vector store with k=12 KNN cosine similarity; sliding-window overlap with section-aware chunking; a metadata pre-filter (year, jurisdiction, patent type, assignee); Cloud Scheduler bi-weekly refresh; and all evaluation fields stored in BigQuery.

Tier 1 / Tier 2 / Tier 3 Patent Inclusion

Drug-name, brand, and Orange Book references resolve to Tier 1; chemical-structure matches to Tier 2; and assignee plus device, formulation, and process signals with a product-specific link to Tier 3.

Multi-Source Data Integration

Espacenet and Google Patents are pre-fetched in parallel, a reverse-engineered IPD fetcher fills the Indian Patent Database gap, and six clinical trial registries are indexed.

Technology Stack

Vertex AI, Gemini, Claude, Agent Development Kit (ADK), Document AI, Vertex AI Vector Search, BigQuery, Cloud Run, Cloud Scheduler, Firestore, Cloud Storage, Secret Manager, ChromaDB (→ AlloyDB planned), FastAPI, and Python.

Key Benefits & Results

Previous: One-shot single-model LLM patent analysis with high hallucination risk.

Our Solution: Gemini extracts and Claude judges with parallel verification.

Result: Every field is cross-checked; failed checks trigger correction with confidence recalculation.

Previous: Tier-1-only patent search that misses contextual matches.

Our Solution: Tier 1 / Tier 2 / Tier 3 taxonomy.

Result: Captures composition-of-matter, formulation, device, process, and dosing patents that assignee-only or direct-mention search misses.

Previous: Single patent source coverage gaps.

Our Solution: Parallel Espacenet + Google Patents pre-fetch plus a reverse-engineered IPD fetcher.

Result: Patent data normally requiring millions in third-party fees, now in-house.

Previous: Tavily API cost for web search.

Our Solution: Migrated to the Vertex AI Google Search tool with domain restriction and keyword match.

Result: Lower cost, better coverage.

Previous: Sequential pipeline runs and slow refresh.

Our Solution: 10 parallel Cloud Run instances with CLOUD_RUN_TASK_INDEX work distribution and Cloud Scheduler automation.

Result: Production-ready bi-weekly refresh across hundreds of drugs.

Previous: Notebook prototypes only (Phase 1).

Our Solution: Production deployment on Cloud Run with BigQuery storage and monitoring.

Result: Workshop patterns scaled into a real production system in Phase 2.

Technical Innovation

Gemini + Claude Multi-Model Judging

Gemini extracts patent data; Claude evaluates and verifies every field in batches of 10 across 8 parallel API calls. Failed checks trigger correction with confidence recalculation — removing single-model hallucination risk entirely.

Tier 1 / Tier 2 / Tier 3 Patent Inclusion Logic

An explicit, codified taxonomy for direct drug, brand, and code mentions, chemical-structure matches, and contextual assignee plus product-type matches. It catches the Tier 3 patents most pipelines miss.

Mandatory Blocking Category Classification

Every patent receives a non-empty classification (composition_of_matter, formulation, device, and more), read directly from the patent claims rather than the title or abstract. This drives downstream dimension routing.

Reverse-Engineered IPD Fetcher

Replaces a third-party service that charges millions for Indian Patent Database fields. Built in-house, it was immediately cost-positive in operation.

Parallel Cloud Run Compute

10 instances per pipeline run with CLOUD_RUN_TASK_INDEX work distribution, exponential-backoff retries for Gemini and Claude rate limits, and incremental processing (insert new, skip unchanged) — production scale, not POC scale.

Wohlig’s Approach

1. Workshop & capability demonstration — a 1-week hands-on build of three modules covering the agentic chatbot, document intelligence, and MCP-server data-lake access.

2. Kickoff & architectural design — defining the multi-pillar Project Cognito architecture, with the IP Pillar selected as Pillar 1.

3. Multi-model pipeline engineering — Gemini + Claude judge integration, ChromaDB retrieval, tiered inclusion logic, and Blocking Category Classification.

4. Multi-source data integration — Espacenet, Google Patents, and the reverse-engineered IPD fetcher; plus ClinicalTrials.gov, PubMed, ChiCTR, EU CTR, CTRI India, and JRCT Japan.

5. Production compute engineering — 10 parallel Cloud Run instances, Cloud Scheduler bi-weekly refresh, and BigQuery evaluation storage.

6. Evaluation framework & continuous iteration — per-dimension metrics (faithfulness, context precision, answer relevancy, cross-dimension coherence), knowledge transfer, the planned AlloyDB migration for production scale, and the upcoming Medical Potential, API Complexity, and Complexity Pillars.

About Dr. Reddy’s Laboratories

Dr. Reddy’s Laboratories Limited (DRL) is a Hyderabad-headquartered global pharmaceutical company specialising in generics, biosimilars, and proprietary products. Its R&D, Manufacturing, Quality, and Biologics functions are leading AI adoption across drug-development decision-making, regulatory document processing, and patent-landscape analysis.

About Wohlig Transformations Pvt. Ltd.

Founded in 2015, Wohlig Transformations specialises in GenAI and DevOps, with 160+ professionals across India and the UK.

Detailed Case Study Presentation: https://www.youtube.com/watch?v=rgk8ev7RBhY

Project Overview

KreditBee partnered with Wohlig Transformations to validate a Google Cloud migration path for its AWS-native application stack — using an AI-assisted development workflow to compress the core migration into 2–3 days and dedicate the remaining engagement to end-to-end testing and standards definition.

KreditBee runs its core application stack on AWS — Lambda (Go), API Gateway, SNS/SQS, S3, and RDS MySQL. As part of its cloud strategy, KreditBee engaged Wohlig to evaluate Google Cloud Platform as a potential migration target. This Google Cloud DAF-funded proof-of-concept migrated the application development environment to GCP, validated the technical feasibility of the equivalent GCP stack, and established the standards that will drive the eventual production migration.

Using an AI-first approach, the engagement delivered 18 Cloud Run Functions, 11 Dockerfiles, 7 Apigee API proxies, and more than 50 Pub/Sub topics while documenting reusable patterns for future production rollout.

The Challenge

Multi-Service AWS Footprint

A core stack spanning Lambda, API Gateway, SNS/SQS, S3, RDS MySQL, Fargate crons, and Secrets Manager required credible Google Cloud equivalents rather than a simple lift-and-shift approach.

Compressed POC Window

The project had a 10-day implementation budget to validate migration feasibility without sacrificing quality, governance, or testing rigor.

Terraform Provider Switch

Existing AWS Terraform configurations needed to be translated to the Google Cloud provider across modules, resources, variables, and deployment workflows.

Delayed Messaging Semantics

AWS SQS delayed-message patterns do not map directly to Pub/Sub and required a different architectural approach to preserve business behavior.

Production-Path Standards

The proof of concept needed to establish reusable naming conventions, infrastructure standards, containerization patterns, and API governance models suitable for production scale.

Key Objectives

• AI-First Migration: Accelerate conversion using AI-assisted development while maintaining human review and testing.

• End-to-End Equivalence: Validate every AWS-to-GCP service mapping through integration testing.

• Terraform-Managed Infrastructure: Deploy all resources as code with no manually configured infrastructure.

• Delayed Messaging Support: Recreate SQS delayed-message behavior using native Google Cloud services.

• Production Standards: Define scalable architecture, deployment, and governance patterns for future migrations.

The Solution: AI-Assisted AWS to GCP Migration

Service Mapping

AWS ServiceGoogle Cloud EquivalentAPI GatewayApigeeLambda (Go)Cloud Run FunctionsSNS / SQSPub/SubS3Cloud Storage (GCS)RDS MySQLSelf-managed MySQL on Compute EngineFargate (Cron Jobs)GKE Pilot / Cloud Run JobsCloudWatch / EventBridgeCloud SchedulerSecrets ManagerSecret ManagerDelayed SQS MessagesCloud Tasks

AI-Assisted Migration Workflow

The migration leveraged AI across three complementary approaches.

Inline IDE Assistance

Used for bulk Lambda-to-Cloud Run Function conversion, replacing AWS SDK integrations with Google Cloud client libraries and generating boilerplate code.

Agentic Migration

Handled Terraform provider translation, complex stateful functions, and messaging pattern conversions from SNS/SQS to Pub/Sub.

Model-Level Pattern Review

Provided reusable migration templates, architecture consistency checks, and QA reviews for AI-generated outputs.

Every workload followed a structured process:

Analyse → Generate → Human Review → Test → Commit

No function was deployed without human validation.

Landing Zone

A production-aligned Google Cloud foundation was established consisting of:

• 1 Organization

• 1 Folder

• 3 Projects (Network Host, Application, Shared Services)

• Shared VPC

• IAM controls

• Organization Policies

• Billing integration

Application Architecture

Apigee serves as the API gateway layer in front of Cloud Run Functions developed in Go.

Pub/Sub and Cloud Tasks manage asynchronous messaging workloads.

Cloud Storage handles object storage requirements.

MySQL runs on Compute Engine behind a bastion host.

Cloud Scheduler orchestrates cron-based processes, while Secret Manager stores application credentials.

CI/CD

Existing GitHub Actions pipelines were migrated to Google Cloud using Workload Identity Federation (WIF) for keyless authentication and Terraform-managed deployments.

Technology Stack

Cloud Run Functions, Apigee, Pub/Sub, Cloud Tasks, Cloud Storage, Cloud Scheduler, Secret Manager, Compute Engine, MySQL, Workload Identity Federation (WIF), Terraform, and GitHub Actions.

Key Benefits & Results

Faster Function Migration

Previous: Multi-week Lambda-by-Lambda migration.

Solution: AI-assisted conversion pipeline with engineering oversight.

Result: 18 Lambda functions migrated to Cloud Run Functions in 2–3 days.

Infrastructure-as-Code Modernization

Previous: AWS-specific Terraform dependencies.

Solution: AI-assisted Terraform provider conversion.

Result: Complete GCP Terraform module suite covering networking, databases, identity, storage, scheduling, messaging, and API management.

API Modernization

Previous: AWS-native API routing.

Solution: Apigee implementation.

Result: 7 API proxies supporting custom domains, CORS policies, and load balancing.

Messaging Modernization

Previous: SNS/SQS architecture.

Solution: Pub/Sub combined with Cloud Tasks.

Result: More than 50 Pub/Sub topics with preserved delayed-message functionality.

Repeatable Deployments

Previous: Console-configured infrastructure.

Solution: Fully Terraform-managed resources.

Result: Version-controlled, repeatable, and code-reviewed deployments.

Production Readiness

Previous: Temporary proof-of-concept implementations.

Solution: Standards-first architecture and documentation.

Result: Production-ready Terraform modules, naming standards, containerization patterns, and Apigee governance guidelines.

Technical Innovation

AI-First Migration Framework

A combination of IDE assistance, agentic workflows, and model-driven review accelerated migration timelines while maintaining engineering quality controls.

Cloud Tasks for Delayed Messaging

Because SQS delayed messages do not directly map to Pub/Sub, Cloud Tasks was introduced alongside Pub/Sub to preserve delayed-delivery behavior without changing business logic.

Workload Identity Federation Everywhere

GitHub Actions and inter-service communication relied on keyless authentication using Workload Identity Federation, eliminating long-lived service account keys.

Inverted Timeline Allocation

Traditional migrations often spend weeks on code conversion and days on validation.

This project completed conversion in 2–3 days and dedicated the majority of the engagement to integration testing and quality assurance.

Production-Scale Standards

All validated patterns—including Terraform modules, naming conventions, containerization approaches, and Apigee configurations—were designed for future production adoption rather than temporary proof-of-concept use.

Wohlig’s Approach

1. Catalogued and classified every Lambda function, trigger, environment variable, and AWS dependency.

2. Established the Google Cloud landing zone, Shared VPC, IAM model, organizational hierarchy, and billing foundation.

3. Used AI-assisted development to migrate 18 critical Lambda functions and convert Terraform configurations to Google Cloud equivalents.

4. Implemented supporting services including Pub/Sub, Cloud Tasks, Apigee, GitHub Actions migration, and Workload Identity Federation.

5. Performed end-to-end validation covering onboarding flows, messaging systems, database connectivity, scheduled jobs, and API routing.

6. Delivered migration runbooks, production recommendations, and knowledge-transfer sessions to the KreditBee engineering team.

About KreditBee

KreditBee, operated by Krazybee Services Limited, is a Bengaluru-based fintech company providing digital lending solutions for salaried and self-employed professionals across India. Founded in 2016, the company is an RBI-registered systemically important Non-Banking Financial Company (NBFC).

About Wohlig Transformations Pvt. Ltd.

Founded in 2015, Wohlig Transformations specializes in GenAI and DevOps, with more than 160 professionals across India and the United Kingdom.

Detailed Case Study : https://www.youtube.com/watch?v=gx-6vMuj4oc

Project Overview

Meesho partnered with Wohlig Transformations to build an end-to-end Document Intelligence platform spanning three key pillars: Google Workspace for document authoring and governance, Gemini Enterprise for conversational access and retrieval, and a custom ADK agent running on Vertex AI Agent Engine to provide depth-aware, fully cited enterprise search.

Meesho operates using a BHAG (Big Hairy Audacious Goals) framework across functions such as Pricing, Monetisation, CPDO, Ranking, and Growth. These teams continuously generate strategic documents including R2R reviews, PFS sessions, working sessions, and KR documents. As the organization scaled, these artifacts became fragmented across individual Google Drive accounts, creating challenges around governance, discoverability, and organizational alignment.

To address this, Wohlig delivered a two-phase platform on Google Cloud.

Phase 1 focused on document governance and ingestion across Google Workspace.

Phase 2 introduced an AI-powered ADK agent integrated into Gemini Enterprise, enabling employees to query organizational knowledge through natural language while preserving document-level access controls.

The Challenge

Inconsistent Document Quality

There was no standardized process to enforce document quality before storage. Teams used different templates, resulting in incomplete and inconsistent documentation.

No Central Repository

BHAG documents were distributed across individual Google Drive accounts, making enterprise-wide knowledge sharing difficult.

Zero AI Discoverability

Employees had no way to search organizational knowledge using natural language.

Incomplete Retrieval Results

The Gemini Enterprise Drive connector returned document snippets rather than complete content, often exposing only a fraction of the available information.

Retrieval API Limitations

The Discovery Engine :answer endpoint permanently rejected custom GCP OAuth implementations, preventing adoption of the intended retrieval architecture.

Service Account Restrictions

Workspace datastore retrieval returned HTTP 403 errors when accessed through service account credentials, requiring an alternative authentication approach.

Key Objectives

• Enforce document quality before documents enter the repository.

• Create a centralized and permission-safe knowledge repository.

• Improve compliance through proactive governance and automated reminders.

• Provide deterministic retrieval depth independent of LLM behavior.

• Ensure complete citation coverage for every answer.

• Eliminate incremental AI inference costs by leveraging existing Gemini Enterprise licenses.

The Solution: Two-Phase Document Intelligence Platform

Phase 1: Document Governance & Ingestion System

Phase 1 introduced ten integrated components governing the entire document lifecycle within Google Workspace.

Custom Template Governance

A Custom Google Template Gallery combined with a Chrome Extension ensured that employees could only access Meesho-approved document templates while hiding Google’s default templates.

8-Rule Audit Engine

An Apps Script Add-on embedded directly into Google Docs validated documents against predefined compliance rules stored in Google Sheets.

The engine:

• Scored each document section from 0–100%

• Evaluated compliance automatically

• Prevented submission until all required sections passed validation

Submit-to-Meesho-Memory Workflow

Documents that passed compliance checks were automatically submitted into the central repository.

The workflow:

• Used the author’s Google Drive token

• Applied repository labels

• Sent confirmation notifications

• Preserved ownership and permissions

Governance & Adoption Automation

Several automated governance mechanisms were introduced:

• In-document nudge notifications

• Meeting invite monitoring

• Daily compliance reminders

• Escalation workflows to managers

• Organization-wide document audits

The backend architecture was later migrated from Apps Script to Cloud Run, with Cloud Scheduler replacing all scheduled triggers.

Keyless Domain-Wide Delegation

JWT signing was performed through the IAM Credentials API, eliminating the need for service account JSON keys.

Phase 2: ADK Agent Inside Gemini Enterprise

The second phase introduced a custom ADK agent registered within Gemini Enterprise Agent Space.

Employees could access organizational knowledge directly from the Gemini Enterprise chat interface they already used daily.

The solution evolved through six architectural versions:

• v1: SequentialAgent baseline

• v2: Parallel per-document extraction

• v3: Wiki-based retrieval

• v4: StreamAssist proof of concept

• v5: Batched parallel StreamAssist architecture

• v6: Production Zero-Vertex Pipeline

Depth-Aware Retrieval Architecture

The production architecture routes queries based on requested depth.

Small Queries

Handled through a single StreamAssist request with suggested follow-up questions.

Medium, Detailed & Exhaustive Queries

A three-stage pipeline executes:

1. Identify

   • Selects the most relevant documents from a candidate pool of up to 91 documents.

   • Generates focused extraction prompts.

2. Batched Parallel StreamAssist

   • Retrieves information from multiple documents simultaneously.

   • Executes 5–10 concurrent retrieval operations for exhaustive searches.

3. Merge

   • Synthesizes results into a single answer.

   • Produces 3–15 citations.

   • Generates follow-up recommendations.

OPT-5 Speculative Prefetch

Candidate retrieval begins in parallel with intent classification, eliminating a complete retrieval round-trip and improving response latency.

Wiki Builder

A Claude-powered wiki generation process consolidates information across multiple documents while preserving inline citations.

Technology Stack

Phase 1

• Cloud Run

• Google Apps Script Add-on

• Chrome Extension (Manifest V3)

• Cloud Scheduler

• Google Drive API

• Google Sheets API

• Drive Labels API

• IAM Credentials API

• Gmail SMTP (Nodemailer)

Phase 2

• Gemini Enterprise

• Gemini Enterprise Agent Space

• StreamAssist API

• Google ADK 1.30.0

• Vertex AI Agent Engine

• Discovery Engine API

• Gemini 3.1 Pro Preview

• Gemini 3 Flash Preview

Security & Access Control

The solution preserves Google Workspace permissions by injecting the end user’s Gemini Enterprise Drive OAuth bearer token into every retrieval request.

This ensures:

• Workspace ACL compliance

• User-level permission enforcement

• No unauthorized document exposure

Keyless Domain-Wide Delegation uses IAM Credentials API signing, eliminating service account key storage entirely.

Key Benefits & Results

Standardized Documentation

Previous: No document template standards.

Solution: Custom Template Gallery and Chrome Extension.

Result: Only approved templates available organization-wide.

Centralized Knowledge Repository

Previous: Documents scattered across personal drives.

Solution: Audit engine and submission workflow.

Result: Centralized repository with compliance enforcement.

Governance Visibility

Previous: Limited awareness of document compliance.

Solution: Automated scans, nudges, and escalation workflows.

Result: Organization-wide compliance monitoring.

Improved Retrieval Quality

Previous: Snippet-only search results.

Solution: StreamAssist retrieval using dataStoreSpecs.

Result: Complete document access with ACL enforcement.

Deterministic Search Depth

Previous: LLM-driven retrieval limitations.

Solution: Python-based parallel retrieval orchestration.

Result: Search depth enforced through code rather than model behavior.

Retrieval API Workaround

Previous: Discovery Engine answer endpoint restrictions.

Solution: StreamAssist API adoption.

Result: Fully functional enterprise retrieval.

Authentication Reliability

Previous: Service account authentication failures.

Solution: Per-user OAuth bearer token injection.

Result: Zero authorization failures.

Full Citation Coverage

Previous: Flash model responses without citations.

Solution: Gemini Pro retrieval with text grounding metadata.

Result: 3–15 citations per answer.

Zero Incremental AI Cost

Previous: Per-token Vertex AI inference costs.

Solution: All retrieval and synthesis workloads executed through Gemini Enterprise licensing.

Result: ₹0 incremental AI cost per query regardless of usage volume.

Technical Innovation

Two-Layer Document Intelligence Architecture

Phase 1 ensures document quality and governance.

Phase 2 makes organizational knowledge discoverable through conversational AI.

AppScript-to-Cloud-Run Migration

The backend was re-platformed during development without service interruption while improving scalability and operational control.

Keyless Domain-Wide Delegation

JWT signing through Google’s IAM Credentials API eliminated the need for service account JSON keys.

StreamAssist-First Retrieval

StreamAssist replaced Discovery Engine’s blocked answer endpoint and became the primary retrieval mechanism.

OPT-5 Speculative Prefetch

Parallel candidate retrieval and intent classification reduced overall query latency.

Zero-Vertex Pipeline

All retrieval, classification, evaluation, and synthesis workloads execute through Gemini Enterprise rather than Vertex LLM agents.

Pro and Flash Model Strategy

Gemini Pro handles retrieval and citation generation.

Gemini Flash handles intent classification for lower latency and cost.

Wohlig’s Approach

1. Defined requirements and designed the governance platform architecture.

2. Built the governance ecosystem, including templates, audit workflows, Chrome extensions, and submission processes.

3. Migrated the backend from Apps Script to Cloud Run.

4. Implemented adoption and governance automation, including nudges, meeting monitoring, escalation workflows, and domain-wide delegation.

5. Completed user acceptance testing and production rollout for Phase 1.

6. Diagnosed limitations of no-code Gemini Enterprise agents, including Discovery Engine restrictions and citation limitations.

7. Iteratively evolved the ADK architecture through six versions before arriving at the production-ready Zero-Vertex Pipeline.

8. Deployed the solution on Vertex AI Agent Engine, registered the ADK agent in Gemini Enterprise Agent Space, and completed organization-wide rollout.

About Meesho

Meesho is India’s leading social commerce platform, enabling millions of small businesses to sell online. Its BHAG-driven operating model generates a large volume of strategic planning documents, making enterprise knowledge discovery and governance critical to operational effectiveness.

About Wohlig Transformations Pvt. Ltd.

Founded in 2015, Wohlig Transformations specializes in GenAI and DevOps solutions, with more than 160 professionals serving clients across India and the United Kingdom.

Detailed Case Study : https://youtu.be/SPQMoSlYox0

Project Overview

Meesho partnered with Wohlig Transformations to build a BigQuery Conversational Analytics agent for its Fulfilment & Experience (FnE) team, achieving 100% effective accuracy on 40 held-out evaluation prompts with every result manually verified row-by-row.

The Fulfilment & Experience (FnE) team at Meesho manages end-to-end operations for India’s largest social-commerce platform using a 10-table mercury dataset in BigQuery, comprising more than 224 columns and 64 defined business metrics.

Wohlig developed a BigQuery Conversational Analytics agent, published as FnE_v2_14Apr in BigQuery Studio’s Agent Catalog, enabling business users to ask natural-language questions and receive SQL queries, BigQuery jobs, result tables, auto-generated visualizations, plain-English insights, and knowledge-source references.

The engagement was executed in two phases. Round 1 achieved 89% effective accuracy across 57 evaluation queries, while Round 2 achieved 100% effective accuracy across 40 held-out evaluation prompts after a dataset quality improvement initiative and complete re-engineering of the knowledge base.

The Challenge

SQL Dependency Across Business Operations

The mercury dataset contained 10 tables, 64 metrics, and 68 business rules. Operational questions around refund clearance, LDR breaches, RTO percentages, NPS, and dispatch performance required analysts to manually write SQL.

Dataset Quality Constraints

The initial implementation reached 89% effective accuracy. Investigation showed that the remaining performance gap stemmed primarily from inconsistencies within the underlying dataset rather than limitations of the conversational analytics agent.

Evaluation Trustworthiness

Enterprise-grade analytics systems cannot rely on LLM-generated evaluation. Every output needed verification at the data level through direct comparison against trusted reference outputs.

Spark-to-BigQuery Conversion Challenges

During migration and reconciliation, 11 Spark normalization issues were discovered and corrected before outputs could be trusted as ground truth.

Overfitting Risk

A critical requirement was ensuring that evaluation prompts remained completely separate from training and verification datasets to avoid artificially inflated results.

Key Objectives

• Enable natural-language access to operational data without requiring SQL expertise.

• Validate outputs through row-by-row and cell-by-cell verification.

• Maintain complete separation between verified training queries and evaluation prompts.

• Ground responses in schema definitions, glossary terms, joins, and business rules.

• Create a repeatable methodology capable of scaling across future datasets.

The Solution: Two-Round Conversational Analytics Program

Round 1: Initial Dataset Evaluation

The first version of the agent was evaluated against 57 reference queries and achieved 89% effective accuracy, consisting of:

• 21 MATCH

• 30 ACCEPTABLE

• 6 genuine failures

Analysis revealed that remaining inaccuracies were driven by dataset quality issues rather than agent logic.

Round 2: FnE Cleaned Dataset

After Meesho engineered a cleaned version of the dataset, Wohlig rebuilt the schema understanding, instructions, glossary, and verified query set.

The updated agent achieved:

• 31 MATCH (77.5%)

• 5 NEAR_MATCH (12.5%)

• 4 ACCEPTABLE (10%)

• 0 NOT_MATCH

Resulting in 100% effective accuracy across all 40 held-out evaluation prompts.

Schema and Knowledge Foundation

The solution incorporated:

• 10 table descriptions

• 8 join relationships

• 88 glossary terms

• 225 documented columns

• 20+ critical field recommendations

The glossary included detailed formula definitions and denominator disambiguation to eliminate ambiguity in business metrics.

Structured Instruction Framework

The agent was governed by:

• 83 MUST/NEVER rules

• 21 instruction categories

• 14 BAD/GOOD examples

• Mandatory filters

• Bucket-metric and date-anchor mappings

This transformed loosely defined business rules into deterministic operational behavior.

Verified Query Set

A set of 73 verified queries was created:

• 57 queries from the MIA Metric List

• 16 supplementary patterns

The verified query set maintained zero overlap with evaluation prompts.

Three-Way Validation Framework

Outputs were validated across:

Spark on Dataproc ↔ BigQuery Conversion ↔ Agent SQL

This process achieved:

• 97/97 functional parity

• 40/40 evaluation queries validated

• 57/57 golden queries validated

Technology Stack

• BigQuery Conversational Analytics API

• BigQuery Studio (Agents Preview)

• Dataplex

• Dataproc

• BigQuery Mercury Dataset

• Gemini

Key Benefits & Results

Reliable Evaluation

Previous: LLM-as-judge evaluation.

Solution: Cell-by-cell manual verification.

Result: Every evaluation verdict was based on actual output data.

Accuracy Improvement

Previous: 89% effective accuracy.

Solution: Cleaned dataset and rebuilt knowledge framework.

Result: 100% effective accuracy on 40 held-out evaluation prompts.

Reference Consistency

Previous: Spark and BigQuery output drift.

Solution: Three-way validation methodology.

Result: 97/97 functional parity and 11 normalization issues identified.

Reduced Hallucinations

Previous: Risk of invented fields and incomplete logic.

Solution: 88-term glossary, 83 instruction rules, and worked examples.

Result: Zero genuine logic failures.

Evaluation Integrity

Previous: Risk of training/evaluation overlap.

Solution: Strict separation between verification and evaluation datasets.

Result: Production-safe and auditable evaluation process.

Improved User Experience

Previous: Manual SQL creation for every business question.

Solution: Conversational Analytics Agent in BigQuery Studio.

Result: SQL generation, execution, visualization, insights, and follow-up recommendations delivered through a single interface.

Technical Innovation

Three-Way Validation Methodology

Every output was validated through direct reconciliation across Spark, BigQuery, and generated SQL before the agent was evaluated.

Manual Row-Level Verification

No LLM-as-judge approach was used. Outputs were evaluated through direct row, column, and cell comparisons. Every NEAR_MATCH and ACCEPTABLE verdict was documented with written justification.

Deterministic Operating Framework

The solution incorporated:

• 88 glossary terms

• 83 MUST/NEVER rules

• 14 worked examples

• Fixed data windows

• 1.5 TB query scan limits

This established a predictable and controlled operational environment.

Repeatable Methodology

The same methodology that identified dataset limitations in Round 1 enabled 100% effective accuracy after data improvements in Round 2, demonstrating repeatability across evolving datasets.

Wohlig’s Approach

1. Collected and audited six input artifacts including tables, metrics, business rules, dimensions, and evaluation prompts.

2. Performed Spark-to-BigQuery conversion with detailed parity validation and correction of normalization issues.

3. Developed schema documentation, glossary definitions, join mappings, and column descriptions.

4. Created structured instruction frameworks, business logic rules, and worked examples.

5. Built a verified query repository with complete separation from evaluation datasets.

6. Trained, evaluated, and published the final agent in BigQuery Studio while delivering full evaluation reports to Meesho and Google for independent review.

About Meesho

Meesho is India’s leading social-commerce platform, enabling millions of small businesses to sell online. Its Fulfilment & Experience (FnE) organization manages the end-to-end customer journey from order placement through delivery, claims, returns, customer support, and post-sale operations.

About Wohlig Transformations Pvt. Ltd.

Founded in 2015, Wohlig Transformations specializes in GenAI and DevOps solutions, with more than 160 professionals serving clients across India and the United Kingdom

.Detailed Case Study : https://www.youtube.com/watch?v=aBwFuxWPGAE

A Wohlig Transformations Whitepaper

Executive Summary

The video tooling category is fragmenting under two simultaneous pressures. Enterprises in regulated sectors cannot send raw footage through third-party SaaS clouds — but their marketing, internal comms, and training teams need to edit video at volume. SaaS products in the creator, ed-tech, and short-video adjacent categories want to embed a video editor as a native feature — but cannot afford the 12–18 month engineering investment to build one.

Wohlig’s Private Video Editor is a deployable platform pattern that resolves both problems. A multi-track, browser-native editor with WebAssembly-based encoding (no server render farm needed for standard formats), deployable inside a customer’s VPC for sovereignty or embeddable inside a SaaS product as a white-label module.

This paper covers the architecture, the two deployment patterns, the extension surface for AI features and high-end rendering, and the engagement model.

1. The Two Problems

Problem A — Sovereignty

For BFSI client testimonial footage, healthcare patient videos, legal recordings, internal HR content, and government communications, sending raw footage to a third-party multi-tenant SaaS is a hard no. Yet the in-house tooling alternative is usually a mix of expensive desktop suites, individually licensed and IT-provisioned.

Problem B — Embedded Capability

SaaS companies in the creator economy, ed-tech, podcast, and short-form video categories want creators to stay in their product. Bouncing out to an external editor breaks the workflow and erodes the product’s defensibility. Building an editor in-house is a major engineering project most product teams cannot fund.

2. Solution Pattern

A browser-native editor architecture that runs entirely client-side for standard formats:

• Multi-track timeline (video, audio, image, text).

• Per-element trim, split, duplicate, layer, position, opacity, z-index, volume.

• Real-time WYSIWYG preview.

• Keyboard-first editing flow.

• Browser-side render via WebAssembly to MP4 up to 1080p.

• No upload to external rendering backend.

Server-side enhancement for cases the browser cannot handle:

• 4K and long-form rendering via a managed render farm.

• Batch export — an ad creative at every aspect ratio in one job.

• AI features — auto-captioning, silence detection, transcript-driven editing, brand-asset enforcement, generative B-roll.

3. Reference Architecture

┌─────────────────────────────────────────────────────┐
│              Customer Domain (white-label)           │
│                                                      │
│   ┌──────────────────────────────────────────────┐   │
│   │  Editor UI  (browser, WASM)                  │   │
│   │   timeline · preview · WASM encoder          │   │
│   └─────────────────┬────────────────────────────┘   │
│                     │                                │
│         standard exports finish here ──── 1080p PCM  │
│                     │                                │
│                     ▼ (only when needed)             │
│   ┌──────────────────────────────────────────────┐   │
│   │  Render Farm  (4K, long-form, batch)         │   │
│   └─────────────────┬────────────────────────────┘   │
│                     │                                │
│                     ▼                                │
│   ┌──────────────────────────────────────────────┐   │
│   │  AI Layer                                    │   │
│   │   captions · silence · transcript · B-roll   │   │
│   └──────────────────────────────────────────────┘   │ 
│                     │                                │
│                     ▼                                │
│   ┌──────────────────────────────────────────────┐   │
│   │  Storage / DAM / CMS  (customer-owned)       │   │
│   └──────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────┘

Identity and authentication via the customer’s existing SSO. All storage in the customer’s tenant.

4. The Two Deployment Patterns

Pattern A — Embedded SaaS Module

A white-label editor module embedded inside a customer’s product (creator economy, ed-tech, podcast, short-video, social schedulers). Branded UI, themed components, hooks into the product’s auth, storage, and content pipeline. Wohlig delivers the foundation, custom features, and integration; the customer owns the deployment and product roadmap.

Pattern B — Private Corporate Editor

A self-hosted editor deployed inside the customer’s VPC or on-prem, gated by SSO. Used by internal marketing, comms, sales enablement, training, and PR teams. Footage and exports stay inside the customer’s perimeter.

5. Extension Surface

The editor foundation is designed to be extended. Common extensions:

Auto-captioning (multi-language) — Marketing, ed-tech, accessibility compliance

Silence detection / auto-cut — Podcast, interview, talk-track production

Transcript-driven editing — Long-form video, podcast, news

Brand-asset enforcement — Marketing, agency white-label

Aspect-ratio presets (Reels, Shorts, OTT, programmatic) — Performance marketing, D2C

Server render farm for 4K / batch — Enterprise marketing, agency

DAM / CMS pipeline — Enterprise content stack integration

Templated intros / outros / lower-thirds — Brand-governed video at scale

6. Outcomes

SaaS video subscription cost reduction — 60–95%

Time-to-edit-and-publish per asset — from days to hours

Footage-leaves-perimeter incidents — zero

In-product editor build effort avoided — 12–18 months of engineering

Embedded editor time-in-product gain — measurable retention lift

7. Engagement Model

Phase A — Foundation (3–4 weeks). Stand up the editor in a staging environment. Configure branding, auth, storage. Validate core editing flows.

Phase B — Customization (4–8 weeks). Add the customer-specific extensions — AI features, render farm, DAM hooks, aspect-ratio presets, templated assets.

Phase C — Launch & operate (ongoing). Deploy to production. Train champions. Wohlig stays as managed service or advisor depending on customer preference.

8. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss embedded video editing or a private corporate editor deployment, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Enterprises that have shipped multiple AI agents into production are now confronting two structural problems. Token spend climbs monotonically because every invocation re-derives the same reasoning. Reliability decays silently because agents break when underlying tools and APIs change, with no automatic remediation loop.

The architectural fix is a Self-Improving Agent Platform — a middleware layer that captures successful workflows as reusable, versioned skills; auto-repairs them when they break; and distributes them across the organization through a governed registry. Industry data on this pattern shows ~46% token reduction and ~4.2x output value per agent invocation.

Wohlig builds this platform for enterprises in BFSI, professional services, BPO/KPO, and document-heavy operations. This paper describes the architecture, the governance model, the integration approach, and the engagement plan.

1. The Two Problems

1.1 Agent Decay

Agents built on raw LLM-plus-tool-call loops are brittle. When a downstream tool updates its API schema, when a regulatory form changes, when an edge case slips through — the agent silently returns wrong answers, or fails outright. Enterprises end up with a growing roster of “broken agents” tickets and the engineering team caught in a maintenance treadmill instead of building new capability.

1.2 Runaway Token Cost

Every agent invocation re-reasons the workflow from scratch. The tax-filing agent thinks through the same logic ten thousand times per month. The compliance agent re-derives the same checks. There is no caching of successful reasoning trajectories — only the much weaker prompt-level cache. Token spend therefore scales linearly with usage and never compresses.

2. Solution Pattern

A middleware layer between agents and tools, with three capabilities:

2.1 Skill Capture

A successful agent run is captured as a named, versioned, structured skill — input schema, tool calls, decision points, output schema, quality metrics. Stored in a registry.

2.2 Skill Retrieval

On a new task, the platform retrieves the most relevant existing skill (hybrid retrieval — lexical, semantic, and LLM-rerank). If a skill matches, the agent executes the skill rather than reasoning from scratch. Token cost drops dramatically.

2.3 Self-Repair

Skill executions are monitored continuously. When a skill fails — schema drift, edge case, tool change — the platform diagnoses the failure, generates a patched version of the skill, validates it against historical executions, and promotes it to the registry. The agent is healed without human intervention.

2.4 Governed Registry

Skills are stored in a registry with public / team / private scopes, lineage DAGs (which skill descended from which), diff-based change tracking, and audit trail. Skills are discoverable across the organization — one team’s win becomes the next team’s starting point.

3. Architecture

┌─────────────────────────────────────────────────────┐
│              Customer Cloud (single tenant)          │
│                                                      │
│  ┌─────────────┐    ┌─────────────────────────────┐  │
│  │  Agents     │    │  Multi-Channel Gateway      │  │
│  │  (existing) │ ◀▶│  WhatsApp · Slack · Teams    │  │
│  └──────┬──────┘    │  Portal · CLI · Desktop     │  │
│         │           └─────────────────────────────┘  │
│         ▼                                            │
│  ┌────────────────────────────────────────────────┐  │
│  │       Skill Engine Middleware                  │  │
│  │  · Retrieval (BM25 + embeddings + rerank)      │  │
│  │  · Execution (run skill or fall back to LLM)   │  │
│  │  · Monitoring (success / failure / drift)      │  │
│  │  · Self-Repair Loop                            │  │
│  └─────────────┬──────────────────────────────────┘  │
│                ▼                                     │
│  ┌────────────────────────────────────────────────┐  │
│  │       Skill Registry                           │  │
│  │  · Versioned skills                            │  │
│  │  · Lineage DAGs                                │  │
│  │  · Quality metrics                             │  │
│  │  · RBAC + scopes                               │  │
│  └─────────────┬──────────────────────────────────┘  │
│                ▼                                     │
│  ┌────────────────────────────────────────────────┐  │
│  │       Tools & Grounding (unified)              │  │
│  │   shell · GUI · MCP · web · enterprise APIs    │  │
│  └────────────────────────────────────────────────┘  │
└─────────────────────────────────────────────────────┘

Deployed inside the customer’s tenancy. Identity through the customer’s SSO. Inference can route to a customer-controlled model endpoint for sensitive workloads.

4. Governance Model

The platform makes AI governance auditable in a way ad-hoc agents cannot:

Skill review — All new skills go through PR-style review before promotion to “team” or “public” scope.

Versioning — Every skill change is a versioned diff with execution-history regression.

Access — RBAC scopes — public / team / private — enforced at the registry.

Audit trail — Every skill execution logged with inputs, outputs, intermediate steps, and cost.

Cost transparency — Per-skill, per-team, per-month cost-per-outcome dashboards.

Failure escalation — Skills that fail repeatedly route to a human review queue rather than auto-patching indefinitely.

5. Outcomes

Token cost reduction — 30–50% (compounds with adoption)

Agent reliability uplift — substantial — silent decay replaced by self-repair

Capability reuse across teams — order-of-magnitude (one skill, many consumers)

Time-to-deploy a new agent capability — from weeks to days

ROI auditability — yes (per-skill cost & outcome metrics)

6. Where It Fits

• Enterprises running 5+ AI agents in production with rising token bills.

• BFSI back-office — KYC, document processing, regulatory filings, dispute handling.

• Professional services — tax, audit, legal, compliance — with high-volume repetitive knowledge work.

• BPO and KPO operations wanting to industrialize AI-assisted delivery.

• Engineering and manufacturing with recurring spec, compliance, and documentation generation.

• Product companies building agent platforms who need a governed skills layer without building it from scratch.

7. Engagement Model

Phase A — Discovery (2 weeks). Inventory existing agents and their failure modes. Map current token spend and reliability metrics. Identify the highest-leverage candidates for skill extraction.

Phase B — Platform deploy (4–6 weeks). Stand up the skill engine, registry, and dashboard inside the customer’s cloud. Wire to existing agent runtimes via the standard protocol layer.

Phase C — Skill migration (8–12 weeks). Wrap existing agents as skills. Establish the review workflow. Onboard 3–5 teams. Begin measuring savings and reliability.

Phase D — Operate (ongoing). Wohlig stays as managed service, hybrid, or advisor.

8. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss your enterprise AI scaling, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Drag-and-drop BI suites — Tableau, Looker, Power BI — were the right architecture for the 2010s. In 2026, they are the source of three expensive problems: per-seat licensing that scales linearly with read access, dashboard sprawl with no enforced single source of truth, and a governance model that does not match the rest of modern software engineering (no diff, no review, no rollback, no audit trail).

Wohlig’s Code-First BI Platform is a deployed-into-your-cloud analytics architecture that replaces these suites with a Markdown + SQL workflow on top of your existing warehouse. KPIs live in version-controlled files. Reviews happen via pull request. Outputs render as a fast static site readable by unlimited users at zero per-seat cost. SOC 2, RBI, HIPAA, and DPDP audit becomes a side effect of using Git.

This paper describes the architecture, the migration approach from existing BI estates, the governance model, and the engagement plan.

1. The Problems with the Current BI Stack

Per-seat cost — ₹15–25 lakh per year for a 100-seat enterprise estate, growing every quarter.

Dashboard sprawl — 60–200 active dashboards typical; majority unused or redundant.

Definition drift — Multiple versions of “active customer” / “net revenue” / “ARR” running in parallel.

No audit trail — Who changed the CAC formula in March, and on whose authority? Usually unknowable.

Slow analytics delivery — Weeks from “we need this report” to “report ships.”

Vendor lock-in — Proprietary file formats; data definitions trapped inside the suite.

2. The Code-First Pattern

A different architectural choice on three axes:

1. Authoring — Markdown files containing fenced SQL blocks and declarative chart components. Stored in Git.

2. Build — A static-site compiler turns the Markdown + SQL into a publishable, fast HTML site. Heavy lifting happens at build time; the runtime is a CDN.

3. Read — Business users open a URL. No client install, no per-seat license, no provisioning ticket.

The query engine pushes down to the warehouse the customer already runs (Snowflake, BigQuery, Redshift, Databricks, Postgres, MySQL, ClickHouse). An embedded columnar engine handles in-page interactive queries on the result sets.

3. Reference Architecture

┌──────────────────────────────────────────────────┐
│        Customer Cloud (GCP / AWS / Azure)        │
│                                                  │
│   ┌──────────────┐                               │
│   │ Git Repo     │ ◀──── Analyst PRs             │
│   │ (.md + SQL)  │                               │
│   └──────┬───────┘                               │
│          │ push                                  │
│          ▼                                       │
│   ┌──────────────┐                               │
│   │ Build & CI   │── lint · dry-run · test       │
│   └──────┬───────┘                               │
│          │ deploy                                │
│          ▼                                       │
│   ┌──────────────┐         ┌────────────────┐    │
│   │ Static Site  │ ◀──────▶│ Warehouse      │    │
│   │ (CDN)        │         │ (Snowflake /   │    │
│   │              │         │  BigQuery /    │    │
│   │  embedded    │         │  Redshift /    │    │
│   │  columnar    │         │  Databricks)   │    │
│   │  engine      │         └────────────────┘    │
│   └──────┬───────┘                               │
│          │ SSO + RBAC + RLS                      │
│          ▼                                       │
│   ┌──────────────┐                               │
│   │ Business     │                               │
│   │ Users        │                               │
│   └──────────────┘                               │
└──────────────────────────────────────────────────┘

Deployment options: any static host inside the customer’s perimeter (Cloud Storage + Cloud CDN, S3 + CloudFront, on-prem nginx). SSO via the customer’s identity provider. Row-level security implemented at the warehouse layer.

4. Governance Model

The code-first pattern moves analytics governance into the same workflow engineering already uses:

• Pull request workflow — every metric change goes through code review.

• Diff-able definitions — what changed in revenue.sql between Q2 and Q3 is one diff.

• Test coverage on critical metrics — automated assertions catch regressions before they ship.

• Lint and dry-run on PR — bad SQL is rejected at the gate.

• Deploy on merge — main branch is production. Hotfix is a PR.

• Audit trail = Git history — SOC 2, RBI, HIPAA, DPDP evidence for analytics changes is the commit log.

For regulated environments, an enterprise tier adds SSO, row-level security, scheduled refresh, and SIEM integration.

5. Migration Approach

Wohlig has a structured migration playbook from existing BI estates.

Step 1 — Inventory. Catalog every active dashboard, its owner, its frequency of use, and its critical metrics.

Step 2 — Consolidate definitions. For every business-critical metric (ARR, CAC, retention, GMV, NPS, etc.), establish one version-controlled SQL definition. Resolve duplicates and conflicts explicitly with the business owner.

Step 3 — Port high-value dashboards. Reproduce the top 20% of dashboards (by usage) on the new platform first.

Step 4 — Run in parallel. Keep the old suite live while the new platform stabilizes. Validate numbers match.

Step 5 — Decommission. Cut over departments to the new platform on a schedule. Reduce the legacy suite to the long-tail dashboards that are not worth porting. Eventually retire it.

Typical timeline: 12–20 weeks from kickoff to legacy decommission for a mid-size enterprise estate.

6. Outcomes

BI license cost reduction — 70–95%

Time-to-ship a new report — from days/weeks to hours

Dashboard count consolidation — 40–70% reduction (kill duplicates and unused)

Audit-ready metric change history — yes (Git native)

Read-seat scalability — unlimited (it’s a website)

Embedded analytics readiness — yes (drop into your product)

7. Engagement Model

Phase A — Discovery & inventory (2–3 weeks). Audit existing estate, define critical metric set, design Git workflow.

Phase B — Foundation deploy (3–4 weeks). Stand up platform in customer cloud, wire to warehouse, configure SSO and RBAC, port first 5 critical reports.

Phase C — Migration (8–16 weeks). Port high-value dashboards. Resolve metric definitions. Train analysts. Run parallel.

Phase D — Decommission (4–8 weeks). Cut over departments. Retire legacy estate. Optimize.

8. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have completed 20+ cloud migrations, shipped 10+ generative-AI solutions in production, and hold 40+ Google Cloud certifications including a Data Analytics specialization. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss your BI consolidation or migration, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Generative AI has reduced the cost of drafting documents by an order of magnitude. The cost of finishing them — making them look on-brand, consistent, and client-ready — has not moved. The result is a new kind of design debt: faster drafts, but inconsistent output, and human cleanup time that erodes the productivity gain.

Wohlig’s Branded AI Document Studio is a deployed-into-your-cloud capability that wraps any AI assistant in a single brand-enforcing constraint layer. Every output is rendered through six standardized document types — one-pager, long document, letter, portfolio, resume, and slides — with palette, typography, logo, spacing, multilingual support, and inline diagram primitives baked in.

This paper covers the problem, the architecture, the integration model with existing AI tools, the brand-DNA configuration, and the engagement plan.

1. The Problem

AI output style drift — Same firm, ten teammates, ten different document looks.

Cleanup tax — 30–60 minutes of human design polish per AI-drafted document.

Brand audit failures — Logos misused, palettes inconsistent, type misapplied.

Localization gap — English-only AI outputs, manual rework for Hindi / regional / CJK.

Knowledge-ops irreproducibility — “What did we send the client in Q2?” has no clean answer.

2. Solution Pattern

A small, deterministic constraint layer that:

1. Lives inside the AI assistant the team already uses.

2. Enforces a single brand specification on every output.

3. Renders the output as a print-ready PDF with no human design step.

4. Supports multiple document archetypes with shared brand DNA.

5. Is owned by the customer, fork-able, and brand-version-controlled.

3. Architecture

┌────────────────────────────────────────────────────┐
│            AI Assistant (existing)                 │
│   (the model your team already uses for drafting)  │
└──────────────────────┬─────────────────────────────┘
                       │  prompt → markdown
                       ▼
┌────────────────────────────────────────────────────┐
│         Branded AI Document Studio (Wohlig)        │
│                                                    │
│   Brand DNA config ──┐                             │
│   ┌──────────────────┴──────────────┐              │
│   │  Six template types             │              │
│   │  · One-pager  · Long doc        │              │
│   │  · Letter     · Portfolio       │              │
│   │  · Resume     · Slides          │              │
│   └────────────────┬────────────────┘              │
│                    ▼                               │
│   Inline SVG diagram primitives                    │
│                    ▼                               │
│   Multilingual font set (EN, HI, regional, CJK)    │
│                    ▼                               │
│   HTML/CSS-to-PDF renderer                         │
└──────────────────────┬─────────────────────────────┘
                       │
                       ▼
              Brand-consistent PDF

The studio is deployed into the customer’s environment — laptop, internal tool, or private cloud, depending on use case — and is fork-able and entirely customer-owned.

4. The Brand DNA Configuration

A single file captures the firm’s visual and editorial identity:

• Palette — primary, secondary, accent, and semantic colors.

• Typography — body, heading, and accent typefaces, with size and weight scales.

• Logo usage — placement rules, clear-space, light/dark variants.

• Spacing & layout — page margins, paragraph rhythm, table styles.

• Voice — short editorial guidelines (tone words, do/don’t list).

• Disclaimers & footers — boilerplate that must appear on regulated document types.

• Languages — which scripts to support and which fonts apply per language.

This single configuration is the source of truth. Updating brand guidelines is a one-line change that propagates to every future output.

5. Document Archetypes Covered

One-pager — Product brief, exec summary, capability sheet

Long document — Proposal, whitepaper, policy memo, RFP response

Letter — Formal correspondence, offer letter, legal letter

Portfolio — Case studies, project showcases, work samples

Resume — Candidate dossier, internal HR profile

Slides — Short, accompanying decks for memos

Custom archetypes (board reports, regulatory filings, audit memos) can be added without breaking the brand layer.

6. Outcomes

Time-per-document polish reduction — 70–90%

Brand-consistency rate of AI-assisted documents — from <50% to >95%

Localization throughput — step change — same pipeline produces EN, HI, regional, CJK

Sales-asset personalization volume — 5–10x more per-account assets

Audit-ready document reproducibility — yes (every output traceable to template + brand version)

7. Where It Fits

• Consulting, advisory, and professional services — proposal and memo factories.

• Investment, legal, and accounting boutiques — branded letters and one-pagers at volume.

• HR-tech and recruiting — candidate dossiers per role.

• In-house enterprise marketing/brand teams — governance over AI-assisted document drafting.

• B2B SaaS sales orgs — per-account proposal generation.

8. Engagement Model

Phase A — Brand DNA capture (1–2 weeks). Extract the firm’s visual, editorial, and regulatory identity into a single configuration file.

Phase B — Studio deployment (2–4 weeks). Stand up the six document templates. Integrate inline diagrams, multilingual support. Wire the studio into the team’s existing AI assistant.

Phase C — Roll out and iterate (ongoing). Train champions in each department. Add custom archetypes. Refine brand DNA as feedback arrives. Wohlig stays as advisor or operator depending on customer preference.

9. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss your branded document automation, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Confixa is Wohlig’s flagship enterprise IT product: a fully autonomous AI IT services firm built as one Master Orchestrator, twelve specialist Domain Agents, and fifty-plus Sub-Agents arranged in a governed hierarchy. It covers software development, testing, DevOps, security, compliance, data engineering, FinOps, vendor management, customer success, and incident operations — all running autonomously, all deployed inside the client’s own Google Cloud project, all audited and governed at the platform level.

Confixa is built for the regulated sectors most software vendors avoid: BFSI, healthcare, and government. It treats compliance, evidence, and audit as first-class system primitives — not bolted-on features.

This whitepaper documents the four-layer architecture, the agent registry, the integration approach with the underlying open-source orchestration platform, the nine critical engineering gaps Wohlig fills with custom services, the architectural risk register, and the phased build roadmap from Phase 0 through Phase 4.

At a Glance:

• Master Orchestrator: 1

• Domain Agents: 12

• Sub-Agents: 50+

• Compliance controls covered: 400+

• Build phases: 5 (Phase 0 through Phase 4)

• Deployment: Inside client’s own GCP project

• Multi-tenant model: One client = one isolated company

• Target verticals: BFSI, healthcare, GovTech

1. Why a New Pattern Is Required

Three structural pressures are simultaneously squeezing regulated enterprise IT in India:

1. Accelerating regulation. DPDP, RBI Master Directions, SEBI CSCRF, IRDAI cyber guidelines, CERT-In incident reporting, and MeitY sector standards all evolve continuously. Manual controls libraries drift out of date within weeks.

2. Persistent talent scarcity. Senior compliance officers, SREs, security engineers, and FinOps specialists are scarce, expensive, and hard to retain. Most enterprises operate 30–40% under-staffed in these functions.

3. Cloud-spend opacity. Unoptimized GCP and AWS estates routinely carry 15–30% recoverable waste — but the analysis requires senior engineering time that is never available.

Point AI tools — a copilot for code, a chatbot for support, a vendor for compliance evidence — do not solve this. The system must be a firm, not a feature: organized into roles, accountable to a governance layer, instrumented end-to-end, and deployed inside the customer’s perimeter.

2. The Four-Layer Architecture

Layer 1 — Client Interaction & Intelligence

Web dashboard for status, compliance posture, approvals queue, agent activity, and FinOps reports. Natural-language WhatsApp commands. Live voice interface for client briefings. Master Orchestrator that decomposes complex client goals into structured subtasks for the right Domain Agents.

Layer 2 — Orchestration Spine

Provided by a self-hosted, MIT-licensed open-source agent orchestration platform. Provides the org chart, ticket system, heartbeat scheduler, governance gates, per-agent budget enforcement, multi-company isolation, runtime context injection, and a pluggable-agent adapter protocol. Configured and deployed by Wohlig on GKE; not coded.

Layer 3 — Infrastructure & Integration Services (Wohlig-built)

The connective tissue that makes the orchestration spine production-ready for enterprise:

• Context API — generates dynamic runtime context for every agent invocation by reading from the Client Profile DB, Evidence Store, current sprint state, and compliance posture cache.

• Real-time Event Bridge — Pub/Sub subscriber that converts events from Grafana, GitHub Actions, vulnerability scanners, ArgoCD, and runtime threat detection into immediate ticket creations. Solves the “heartbeats are too slow for incidents” failure mode.

• Approval Workflow Engine — parses approval/rejection signals from email, WhatsApp, and Slack and translates them into ticket status updates. Configurable SLA with escalation to backup approvers.

• Secrets Bridge — synchronizes credentials between GCP Secret Manager and the agent layer. No long-lived credentials in agent config. Rotation triggers automatic config update.

• Agent Adapter Framework — standardized Python/Node.js wrapper library. Any new agent registers as a pluggable-agent endpoint via this framework. Handles heartbeat acknowledgement, task checkout, state persistence, result reporting, and PII scrubbing.

• Agent Observatory — measures task success rate, escalation rate, hallucination flag rate, MTTD/MTTR contribution, cost-per-outcome. Publishes a weekly Agent Health Report.

Layer 4 — Capability Services (Wohlig-built — the product)

The actual IT services Confixa delivers:

• Compliance Engine — controls library across DPDP, RBI, SEBI, IRDAI, ISO 27001, SOC 2, GDPR, HIPAA, PCI-DSS. 400+ controls, automated check scripts, evidence collection, gap assessment, regulatory change processing.

• Evidence Store — Cloud SQL evidence metadata + GCS artifact store. Schema: evidence_id, client_id, framework, control_id, artifact_type, collected_at, artifact_url, status. Source of truth for audit packs.

• Audit Pack Generator — assembles formatted audit packages from evidence: PDF executive summary, per-control evidence folders, risk register, narrative auditor responses. Sub-4-hour SLA from trigger.

• Regulatory Intelligence Feed — monitors RBI, SEBI, IRDAI, MCA, CERT-In, MeitY portals on daily heartbeat. New publication → classification → impact-assessment ticket within 24 hours.

• Security Toolchain — adapters for SAST, SCA, DAST, runtime threat detection, secrets scanning. Normalized vulnerability schema. SLA-tracked remediation tickets.

• Development Services — PR creation and review, multi-language, tech debt tracker, dependency management.

• Data Platform — BigQuery, dbt pipelines, Looker Studio dashboards, PII masking, natural-language-to-SQL service.

• FinOps Services — GCP Cost Explorer integration, rightsizing analysis, committed-use discount optimizer, monthly cost reports.

• Client Onboarding Agent — automated workflow that provisions a new client company, registers all 12 Domain Agents, generates initial SKILLS context files, runs the first compliance baseline, and delivers a Day 1 report — in under 4 hours from credentials grant.

3. Agent Registry

Master Orchestrator (Tier 1)

Receives client goal tickets, decomposes into subtasks, creates child tickets for Domain Agents. Wakes on ticket assignment.

Domain Agents (Tier 2)

Requirements & Planning — Heartbeat: 8h — Owns: Backlog, sprints, user-story extraction

Development — Heartbeat: On ticket — Owns: PR creation, peer-review pipeline

Testing & QA — Heartbeat: On ticket + 12h — Owns: Unit, integration, E2E generation

DevOps & Deployment — Heartbeat: 4h — Owns: CI/CD, GitOps, IaC with approval gate

Security & Vulnerability — Heartbeat: 2h — Owns: SAST, SCA, DAST, secrets, runtime

Compliance & Audit — Heartbeat: 6h — Owns: Controls library, evidence, gap assessment

Data & Analytics — Heartbeat: 12h — Owns: Datasets, dashboards, PII classification

Vendor & Procurement — Heartbeat: 24h — Owns: RFPs, contracts, vendor SLA tracking

Customer Success & Demo — Heartbeat: On ticket — Owns: Demo provisioning, voice briefings

FinOps & Cost — Heartbeat: 24h — Owns: Spend, rightsizing, anomaly tracking

Incident & Operations — Heartbeat: 30m — Owns: Alerts, runbooks, RCA generation

Documentation — Heartbeat: Weekly — Owns: Doc coverage, stale doc register, API specs

Sub-Agents (Tier 3 — sample)

Code Review Bot, Secrets Scanner Bot, Evidence Collector Bot, Audit Doc Bot, Policy Checker Bot, Rightsizing Bot, RCA Bot, User Story Bot, Demo Provisioning Bot, Threat Model Bot — fifty-plus in total.

4. Coverage Analysis

A formal coverage analysis was performed against the underlying orchestration platform. Of the 27 Confixa requirements:

• 10 covered fully by the orchestration platform — no Wohlig build needed; configure and deploy.

• 4 covered partially — orchestration platform provides scaffolding; Wohlig builds the capability layer on top.

• 13 require entirely new Wohlig builds — the differentiated capability that makes Confixa worth paying for.

The platform contributes ~37% of operational primitives. The remaining 63% — compliance engine, evidence store, audit pack generator, regulatory intelligence feed, security toolchain adapters, real-time event bridge, agent observatory, data platform, vendor agent, FinOps, customer-success agent, client-onboarding automation, GKE-on-Cloud-SQL deployment — is the Confixa product.

5. The Nine Engineering Gaps

The integration analysis surfaced nine critical gaps that must be filled by Wohlig before any Domain Agent can serve a regulated client:

1. No dynamic context injection (Context API) — Phase 0

2. No real-time event bridge (incidents stall on heartbeat) — Phase 0

3. No approval workflow loop (approval gates dead-end) — Phase 1

4. No evidence store or audit pack generator — Phase 1

5. No regulatory intelligence feed — Phase 1

6. No agent performance monitor (KPIs unmeasurable) — Phase 2

7. No GCP Secret Manager integration (BFSI blocker) — Phase 0

8. Orchestration platform not GKE-deployable out of box — Phase 0

9. No client-onboarding automation (manual = unscalable) — Phase 1

Each is addressed in the build roadmap below.

6. Architectural Risks (Selected)

Orchestration platform breaking changes — Severity: High Mitigation: Wohlig fork pinned on Artifact Registry; thin abstraction layer for all platform calls.

Sub-agent explosion (2,500+ registrations at scale) — Severity: Medium Mitigation: Lazy registration; sub-agents register only on first activation per client.

Cloud SQL connection pool exhaustion — Severity: High Mitigation: PgBouncer pooler; HPA at 70% pool utilization.

LLM API rate limits during peak heartbeat cycles — Severity: Medium Mitigation: Priority queuing — Incidents > Security > DevOps > Compliance > Analytics; tiered model usage; prompt caching.

Heartbeats too slow for real-time incidents — Severity: Critical Mitigation: Solved by Real-time Event Bridge — non-negotiable Phase 0.

DPDP data localization for client data in model APIs — Severity: High (BFSI) Mitigation: PII scrubbing middleware in Agent Adapter Framework — DLP inspection before any model call; placeholder substitution; full audit.

7. Phased Build Roadmap

Phase 0 — Foundation (Weeks 1–6)

Container the orchestration platform for GKE. Cloud SQL, Workload Identity, Secret Manager integration, GCS bucket, Agent Adapter Framework, Context API v1, Real-time Event Bridge, all 12 Domain Agents registered as pluggable-agent skeletons. All P0. Blockers for production.

Phase 1 — Autonomous DevOps Core (Weeks 7–42)

Approval Workflow Engine. Client Onboarding Agent. Regulatory Intelligence Feed. Evidence Store. DevOps Agent (CI/CD, ArgoCD, Terraform). Security Agent (SAST, SCA, secrets scanning). Compliance Agent (controls library v1 — CERT-In, ISO 27001, DPDP, 100 controls; 6-hour automated checks; consent management). WhatsApp natural-language commands. FinOps v1.

Phase 2 — Full SDLC Autonomy (Weeks 43–78)

Audit Pack Generator. Agent Observatory. Development Agent (PR creation, peer-review pipeline). Testing Agent (unit, integration, E2E from user stories, k6 performance). Requirements Agent (multi-channel ingestion, PRD/User Story auto-generation). Customer Success Agent (Live voice adapter, demo provisioning). PII scrubbing middleware. DAST adapter. Falco runtime. RBI Digital Lending guidelines. SEBI CSCRF gap assessment. SOC 2 Type I readiness. Auditor Query Portal. Documentation Agent.

Phase 3 — Enterprise & Regulatory Scale (Weeks 79–118)

Data & Analytics Agent (BigQuery, NL-to-SQL, dbt, Looker Studio). Vendor Agent (RFP generation, contract risk analysis). BFSI Industry Pack (RBI + SEBI + IRDAI + DPDP integrated controls). Healthcare Industry Pack (HIPAA + DPDP + clinical data). GovTech Industry Pack (MeitY + CMMI + NIC). GDDR. SOC 2 Type II continuous. PCI DSS v4.0. Industry pack templates. Adapter abstraction hardening. Agent pool scaling tests (500 agents × 20 clients). FinOps v2 (auto-implement low-risk savings).

Phase 4 — Fully Autonomous IT Firm (Weeks 119–156)

Strategic Planning Agent. Capacity & Skills Agent. Multi-client scale (50+ concurrent companies). Outcome-based billing engine. White-label mode for IT services firms. Agent self-improvement loop. Cross-client anonymized knowledge transfer. GCP Marketplace listing.

8. Compliance Model

Compliance is a first-class primitive, not a bolt-on. Three principles:

1. Continuous, not point-in-time. The Compliance Agent runs automated control checks on a 6-hour heartbeat. Posture is always current; auditor queries do not require a fire drill.

2. Evidence-first. Every check writes evidence to the Evidence Store, tagged to the control ID. Audit packs assemble in under four hours from trigger.

3. Regulator-aware. The Regulatory Intelligence Feed converts each new RBI / SEBI / IRDAI / CERT-In / MeitY publication into a ticket within 24 hours. The controls library does not drift.

Frameworks covered (Phase 1 → Phase 3): DPDP, RBI Master Directions, SEBI CSCRF, IRDAI cyber guidelines, CERT-In, ISO 27001, SOC 2 Type I and II, GDPR, HIPAA, PCI DSS v4.0, MeitY, NIC standards.

9. Commercial Logic

Confixa is priced as a replacement for an IT services relationship, not as software seats. The economics:

• Reference baseline: mid-tier managed-service contracts for regulated enterprise IT run ₹2–8 crore per year for the scope Confixa covers.

• Confixa value drivers: 24×7 operation, audit-ready evidence on demand, recovered cloud-spend savings (typically 15–30% of GCP bill), zero attrition risk, full data sovereignty.

Pricing is set per phase and per industry pack. Phase 4 unlocks outcome-based billing for enterprise accounts.

10. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have completed 20+ cloud migrations, shipped 10+ generative-AI solutions in production, and hold 40+ Google Cloud certifications including a Data Analytics specialization. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss Confixa for your enterprise or to evaluate a pilot, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Pocket Studio 2.0 is Wohlig’s flagship AI product: a fully autonomous digital agency built as eighteen specialist AI agents arranged in a five-tier corporate hierarchy. It covers every discipline a top global agency offers — strategy, research, creative production (copy, image, video, audio), SEO, paid media, PR, influencer marketing, community management, landing pages, CRO, marketing automation, analytics, and client servicing — and it runs them autonomously, end-to-end, from a single client brief to a live, compliant, optimized, continuously improving campaign.

This whitepaper documents the architecture, the governance model, the agent capability profiles at a summary level, the campaign execution flow, the technology stack, the phased build roadmap, and the go-to-market strategy.

At a Glance:

• Specialist agents: 18

• Org-chart tiers: 5 (Board → CEO → C-Suite → VP → Specialist)

• Agency capability coverage: 100% (zero gaps vs. top global agencies)

• Build phases: 3, total ~12 months

• System cost per client: very low — see commercial section

• Infrastructure: Google Cloud Platform (data sovereignty)

• Client pricing: ₹1.5L (Phase 1) → ₹5L+ (Phase 3) per month

1. Why a New Architecture is Required

The current AI marketing landscape is a graveyard of point tools. One vendor automates copy. Another automates images. A third schedules posts. None connect. None enforce brand. None audit themselves. None respect regulatory boundaries for BFSI, pharma, education, or real estate. None survive a serious enterprise procurement review.

A real digital agency is not a sequence of tools. It is a company — with strategists, creatives, performance leads, account managers, and a quality and compliance discipline that runs through every output. To replace it, you must architect a company, not a feature set.

That is what Pocket Studio 2.0 is.

2. The Five-Tier Hierarchy

Every agent has a title, a manager, reporting lines, a monthly budget, and a heartbeat schedule. The hierarchy is enforced at the platform level — agents can only delegate downward and escalate upward.

Tier 0 — Board (Human Governance Layer). Wohlig account team plus client stakeholders. Approves strategy, overrides agents, sets budgets, reviews QBRs. Notified via WhatsApp for every approval gate.

Tier 1 — CEO Agent (Maestro). Receives briefs, parses them, opens sub-tickets to department heads, manages revisions, escalates blockers to the board, routes deliverables through compliance and client approval.

Tier 2 — C-Suite (3 agents).

• Strategist — Chief Strategy Officer.

• Sentinel — Chief Compliance Officer.

• Diplomat — Chief Client Officer.

Tier 3 — VP / Director Layer (8 agents). Functional heads for research, brand monitoring, performance analytics, infrastructure, paid media, PR, marketing automation, and community.

Tier 4 — Specialist / IC Layer (6 agents). The execution layer — copywriting, art direction, video production, web engineering, influencer management, and community engagement.

3. Agent Capability Coverage

The eighteen agents collectively cover the full service catalog of a top global agency. A summary view:

Strategy & planning — Strategist + Oracle-Market + Oracle-Pulse

Creative — copy, design, video — Scribe + Visualist + Director + Builder

SEO & performance marketing — Amplifier (Google, Meta, LinkedIn, programmatic, ASO)

PR & earned media — Publicist (wire, media relations, crisis)

Influencer marketing — Influencer (discovery, outreach, UGC pipeline)

Community & social management — Engager (comments, DMs, reviews, communities)

Marketing automation & CRM — Automator + Nexus (scoring, lifecycle, ABM)

Web production & CRO — Builder (pages, technical SEO, CRO, Core Web Vitals)

Analytics & reporting — Analyst + Diplomat (daily reports, QBR decks)

Account management — Maestro + Diplomat + Sentinel (orchestration, compliance, comms)

A formal agency capability gap audit confirmed zero gaps across content & copywriting, creative production, distribution & publishing, paid media & performance, web & SEO, intelligence & research, community & engagement, PR & communications, marketing automation & CRM, analytics & reporting, and client servicing & account management.

4. The Orchestration Layer

Pocket Studio 2.0 runs on a self-hosted, open-source orchestration platform that models the system as a company rather than a workflow. This layer provides:

• Org chart enforcement — agents cannot operate outside their authority.

• Heartbeat scheduling — agents wake on the right cadence for their job (continuous, reactive, or batched).

• Ticket system — every task is a ticket with full goal ancestry, parent-child relationships, message thread, and tool-call trace.

• Budget control — each agent has a hard monthly token budget enforced automatically. No runaway spend.

• Governance — the board can pause, override, reassign, or terminate any agent at any time. Approval gates are enforced at the platform level, not through agent logic.

• Goal alignment — every action carries lineage back to the original client brief. Agents always know why they are doing what they are doing.

• Multi-company — one platform install runs many client companies in parallel with complete data isolation. Wohlig operates a single control plane for the entire portfolio.

• Skill discovery — runtime context (brand guidelines, campaign history, compliance rulesets, client preferences) discoverable on demand by agents — no need to inject into every prompt.

Without this layer, Wohlig’s engineering team would have to build — from scratch — a stateful campaign management system, agent routing engine, budget enforcement layer, audit logging system, multi-tenant isolation, board approval workflow, and a heartbeat scheduler. The orchestration platform provides all of this off the shelf, saving 4–6 months of senior engineering time in Phase 1 alone.

5. Heartbeat Schedules

Three categories of agent wakeup logic, optimized for cost and responsiveness:

• Continuous agents wake every 2–8 hours for monitoring and publishing tasks. Brand-monitoring runs every 4 hours. Publishing queue checks every 2 hours. Comment moderation every 2 hours. Lead score evaluation every 6 hours. Paid media review every 8 hours.

• Reactive agents wake only when a ticket is assigned — zero idle activity. CEO, Strategist, Compliance, Copywriter, Visualist, Director, Influencer, and Client Officer.

• Scheduled agents wake on fixed daily or weekly cadence for batch analytics — daily 6 AM analytics aggregation, weekly Monday 7 AM CRO and heatmap analysis.

This pattern keeps token spend predictable and gross margins very strong even at low client price points.

6. End-to-End Campaign Execution Flow

Brief received → Maestro parses, opens ticket #001
   │
   ▼
Oracle-Market → 20–30 min pre-campaign intelligence (competitive,
                 audience, market, channel)
   │
   ▼
Strategist → builds full campaign plan (channel mix, calendar, KPIs,
              creative briefs, influencer plan, ABM list, automation
              flows)
   │
   ▼
[ Board Approval Gate 1 — Strategy ]   ← enforced at platform level
   │
   ▼
Scribe / Visualist / Director / Builder / Influencer
              → parallel creative production
   │
   ▼
Sentinel → compliance screen on every asset
              (brand DNA + regulatory ruleset)
   │
   ▼
[ Board Approval Gate 2 — Creative review ]   ← enforced at platform
   │
   ▼
Broadcaster + Amplifier + Builder → live distribution
              (social, paid, programmatic, page launch, GTM)
   │
   ▼
Analyst + Oracle-Pulse → continuous performance monitoring,
                          creative fatigue, brand sentiment
   │
   ▼
Diplomat → weekly client report + monthly executive summary +
            quarterly business review deck
   │
   ▼
Analyst → historical learning brief stored in client knowledge base
            and surfaced to Strategist on the next brief

A real example: a single client brief for a course launch campaign generates 22 tickets across 18 agents — every one logged, every one traceable from any leaf back to the original brief.

7. Compliance — Built In, Not Bolted On

Sentinel screens every asset before it advances. Regulatory rulesets covered:

• India — ASCI (general advertising), RBI / SEBI / IRDAI (BFSI), Drugs & Magic Remedies Act and FSSAI (pharma & food), RERA (real estate), Cable TV Networks Act, plus state-level regulations (Kerala lottery, Maharashtra RERA, alcohol advertising rules).

• International — UK ASA, UAE NMC, US FTC influencer disclosure rules.

Every influencer post is screened for the correct disclosure format per market — #ad in India, #ad in UK, #sponsored / #ad in US.

Any asset that fails compliance is rejected automatically with an annotated reason and a revision ticket opened to the originating agent. Sentinel runs on every asset — no exceptions. Compliance clearance certificates are issued in the ticket system.

This is the single capability that closes enterprise deals in regulated sectors.

8. Analytics & Intelligence

Three agents own the complete analytics picture:

• Oracle-Market — pre-campaign intelligence (competitive, audience, market, channel) — runs per brief and on monthly refresh.

• Oracle-Pulse — always-on monitoring (brand sentiment, competitor ads, viral trends, review platforms) — 4-hour heartbeat.

• Analyst — post-campaign performance (campaign metrics, creative fatigue, SEO, attribution & ROI, predictive analytics, historical learning brief) — daily, weekly, monthly, and per-campaign-end.

The historical learning brief is the system’s compounding advantage — every campaign makes the next one measurably smarter for that client.

9. Web, SEO & CRO

The Builder agent owns every functional web output: landing page production (HTML/CSS, Webflow, Framer, Unbounce, WordPress exports), A/B variants, technical SEO on every page (meta, schema, canonicals, hreflang, internal linking), Core Web Vitals audit (LCP < 2.5s, CLS < 0.1, INP < 200ms), CRO monitoring (heatmap, scroll, form abandonment, A/B tests), full GTM container setup, and HTML5 animated banner production.

Every landing page goes through a structured 9-step pipeline across 6 agents with two human approval gates — brief, research, brand & compliance load, copy, visuals, page assembly, compliance screen, client approval, live with continuous CRO monitoring.

10. Phased Build Roadmap

Phase 1 — Months 1–4 — Core Intelligence & Orchestration. Orchestration platform on GCP. 8 foundational agents (Maestro, Oracle-Market, Oracle-Pulse, Sentinel, Scribe, Diplomat, Strategist, Builder basic). Brief-to-brand-compliant-copy-and-creative delivered via WhatsApp approval. Priced at ₹1.5–2L per client per month. Two or three pilot clients fund Phase 2.

Phase 2 — Months 5–8 — Full Creative Production & Distribution. All creative production agents online — Visualist, Director, Builder full, Broadcaster, Amplifier, Analyst, Influencer (basic), Client Portal v2. End-to-end autonomous. Priced at ₹3.5–5L per client per month.

Phase 3 — Months 9–12 — Scale, Automation, Community & PR. Influencer full, Engager, Publicist, Automator. All 11 agent extensions. Programmatic via DV360 and The Trade Desk. Enterprise CRM and CDP integrations. White-label mode. GCP Marketplace package. Priced at ₹5L+ or outcome-based for enterprise.

Each phase ships something usable and billable while the next layer builds on top.

11. Technology Stack (Summary)

• Infrastructure — Google Cloud Platform: GKE for agent containers, Cloud Run for portal endpoints, Firestore for state, BigQuery for analytics, Cloud Storage for assets, Pub/Sub for events, Cloud Tasks for heartbeats, Looker Studio for client dashboards, Secret Manager, Cloud Armor, IAP.

• Creative production — best-in-class image, video, and voice generation tools, selected per agent for the job.

• Research & comms APIs — Tavily, Perplexity, SEMrush, SerpAPI, Brandwatch, Sprinklr, HypeAuditor, Modash, Muck Rack, Cision, PR Newswire, Meta Graph API, WhatsApp Business API, Google Ads API, LinkedIn Marketing API, DV360 + The Trade Desk, HubSpot, Salesforce Marketing Cloud, Marketo, Twilio, Kaleyra, SendGrid.

• Multi-tenant on GCP — each client company is a separate namespace with isolated org chart, brand database, agent budgets, and audit logs.

12. Team Structure

Total engineering team at full build: 14 people.

• Phase 1 (6 people) — 1 AI/ML engineer, 2 backend engineers, 1 frontend engineer, 1 product manager, plus shared leadership.

• Phase 2 additions (4 people) — creative tech engineer, integrations engineer, data engineer, web engineer.

• Phase 3 additions (4 people) — PR & community engineer, CRM & automation engineer, DevOps / platform engineer, enterprise solutions engineer.

13. Go-to-Market and Commercial Logic

Pocket Studio 2.0 is not priced as software seats. It is priced as a replacement for an agency relationship.

Phase 1 — Primary segment: GCP customers spending ₹15–25L/month on external agencies — Pricing: ₹1.5–2L / month (Starter)

Phase 2 — Primary segment: Fast-growth D2C, EdTech, fintech needing full creative & multi-channel — Pricing: ₹3.5–5L / month (Growth)

Phase 3 — Primary segment: Enterprise multi-brand replacing agency relationships — Pricing: ₹5L+ / month or outcome-based

A mid-size agency retainer is ₹15–25L/month for comparable output. Pocket Studio 2.0 at ₹3.5–5L/month — with 24×7 operation, audit-grade governance, regulatory compliance, and full data sovereignty — closes the ROI conversation for almost every prospect in the target segment.

14. Competitive Moat

Three structural advantages:

1. The compliance engine. No competitor screens for SEBI, IRDAI, FSSAI, RERA, ASCI, UK ASA, UAE NMC, and US FTC simultaneously. Building this takes 12–18 months and deep regulatory expertise per sector. It is mandatory for BFSI and pharma — India’s largest marketing spenders.

2. Full data sovereignty on GCP. Every competitor sends client data to their own SaaS cloud. Pocket Studio 2.0 lives in the client’s own GCP project. For banks, insurers, pharma, and government-adjacent organizations, this is non-negotiable.

3. End-to-end campaign autonomy. Every competitor automates a fragment. Pocket Studio 2.0 automates the entire agency relationship — brief to live campaign to performance report to next brief — with compliance built into every step.

Multi-year head start, on a moat that compounds with every client campaign.

15. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications including a Data Analytics specialization. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss Pocket Studio 2.0 for your brand or agency, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Email remains the highest-volume, highest-cost knowledge-work surface in modern enterprises. A relationship manager, support agent, or client-services lead spends 30–50% of their day reading and composing messages that are structurally similar to messages they have written hundreds of times before. Off-the-shelf AI assistants do not solve this problem at the enterprise level: they live outside the firm’s perimeter, lack access to the firm’s customer and operational data, and are not safe for regulated correspondence.

Wohlig’s Agentic Inbox is a private, deployed-in-customer-cloud email and messaging platform that auto-drafts a reply for every inbound message, grounded in the firm’s real systems (CRM, ticketing, KB, domain tools), with a human-in-the-loop approval step. It compresses first-response time, raises throughput per agent, enforces tone and policy, and keeps all data inside the customer’s perimeter.

This paper outlines the problem, the architecture, the integration model, the security posture, and the engagement plan.

1. The Problem

Volume — Senior knowledge workers receive 60–150 emails per day.

Repetition — A large share of replies are structurally similar.

Context-switch cost — Each email forces a customer/case/policy load into working memory.

Tone and policy variance — Replies vary by author, day, and mood.

SLA pressure — First-response targets are missed during peak windows.

Compliance risk — Sensitive content cannot be routed through public AI SaaS.

2. The Pattern: A Private Agentic Inbox

2.1 Auto-draft on inbound

Every inbound email triggers the agent to compose a candidate reply before a human opens the message. The draft is composed using:

• The full thread context.

• A configurable per-mailbox or per-user system prompt encoding tone, role, and policy.

• Tool calls into the firm’s real systems (CRM, ticketing, KB, domain tools) for grounded facts.

2.2 Human-in-the-loop confirm

The human reviewer approves, edits, or overrides every outbound message. The agent never sends without explicit human action — a critical control for regulated and client-facing contexts.

2.3 Per-mailbox isolation

Each user or shared mailbox has its own isolated state, chat history, and document storage. There is no cross-tenant data co-mingling.

2.4 Tool calling, not hallucination

The agent answers from real data — pulled at draft-time from CRM, ticketing, KB, calendar, and domain systems — not from the model’s training. This is the line between “AI assistant” and “AI assistant your CISO will approve.”

2.5 Single-tenant, customer-cloud deployment

Mailboxes, attachments, chat history, and inferencing all live inside the customer’s cloud tenancy and own domain. Authentication is enforced through the customer’s existing SSO.

3. Reference Architecture

        ┌───────────────────────────────────────────────┐
        │       Customer Domain (mail.acme.com)         │
        │                                               │
        │   ┌───────────┐  ┌────────────────────────┐   │
        │   │ Mail Ingress││  Agentic Inbox UI      │   │
        │   └─────┬──────┘ └────────┬───────────────┘   │
        │         │                 │                   │
        │         ▼                 ▼                   │
        │   ┌──────────────────────────────────┐        │
        │   │      Per-Mailbox Workers         │        │
        │   │  (state, chat history, drafts)   │        │
        │   └────┬─────────────────┬───────────┘        │
        │        │                 │                    │
        │  ┌─────▼──────┐    ┌─────▼─────────┐          │
        │  │ AI Agent   │    │ Attachment    │          │
        │  │ (LLM +     │    │ Object Store  │          │
        │  │  Tools)    │    │               │          │
        │  └─────┬──────┘    └───────────────┘          │
        │        │                                      │
        └────────┼──────────────────────────────────────┘
                 ▼
        ┌──────────────────────────────────┐
        │     Customer Systems (read-only   │
        │     or scoped write):             │
        │     CRM · Ticketing · KB · Cal    │
        │     Domain tools (PMS, EHR, etc.) │
        └──────────────────────────────────┘

Edge-native deployment options for low operational overhead, or equivalent GCP/AWS-native components for customers standardized on a hyperscaler. Per-tenant state isolation through serverless workers with embedded SQL state stores. SSO via the customer’s existing identity provider (Okta, Azure AD, or equivalent).

4. Security and Compliance Posture

The agentic inbox is designed for regulated environments. Key controls:

• Per-mailbox state isolation. No cross-user data access.

• SSO / Zero-Trust authentication on every request.

• All data in customer perimeter. Mail, attachments, agent state.

• Scoped tool permissions. Each tool call uses least-privilege credentials with a per-tenant audit log.

• Human-in-the-loop send. No autonomous outbound by default.

• PII handling and redaction policies configurable per mailbox.

• Audit trail. Every draft, edit, and send recorded with reviewer identity and timestamp.

• Configurable retention to satisfy regulatory record-keeping.

5. Outcomes

First-response time reduction — 40–80%

Knowledge-worker throughput increase — 2–4x

Tier-1 inquiry deflection — 30–60%

Tone/policy variance reduction — substantial; measurable through QA sampling

Cost-per-contact reduction — 25–50%

6. Where It Fits

• BFSI — relationship management, KYC correspondence, advisor email.

• Legal and accounting firms — privileged client mail, document requests, scheduling.

• B2B SaaS — customer-success, support, renewals, AE inboxes.

• Healthcare and clinics — appointment, billing, records.

• Internal shared mailboxes — sales@, support@, careers@, ops@.

7. Engagement Model

Phase A — Foundation (3–4 weeks). Stand up the inbox stack in the customer’s cloud. Connect SSO, mail routing, and a first system of record (e.g. CRM). Pilot with one team and one mailbox.

Phase B — Integration (4–8 weeks). Wire in additional systems (ticketing, KB, calendar, domain tools). Configure per-mailbox tone and policy. Roll out to additional teams.

Phase C — Operate (ongoing). Wohlig operates, advises, or hands over depending on customer preference. Continuous evaluation, guardrail tuning, and tool expansion.

8. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss your customer-correspondence transformation, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

The buy-side data and workflow stack — Bloomberg, Refinitiv, Capital IQ on top, FactSet in the middle, scattered Python notebooks at the bottom — is structurally a 1990s product retrofitted with cloud features. It is expensive (₹15–30 lakh per seat for the senior tier), siloed, and fundamentally a “dashboard for humans to do the work themselves.”

A new generation of multi-agent AI architectures changes what is possible. By combining a Bloomberg-style desktop terminal, a fleet of specialized analyst agents, 100+ real data connectors, broker connectivity, and audit-grade logging, Wohlig builds AI-native investment desks that compress days of analyst work into minutes while running entirely inside the customer’s perimeter.

This paper outlines the capability, architecture, governance layer required to operate it on real money, and engagement model.

1. The Problem with the Buy-Side Stack Today

Vendor cost — Bloomberg / Refinitiv / Capital IQ at ₹15–30L/seat per year.

Workflow fragmentation — 5–10 tools per analyst — research, screener, risk, OMS, journals.

Research bottleneck — Analyst teams cover a fraction of the universe; ideas die in the queue.

Discretionary risk discipline — Sizing and exposure checks skipped under time pressure.

Idea-to-execution lag — Days from thesis to sized order; missed entry windows.

Audit deficit — No comprehensive record of why a position was taken.

2. The Pattern: An AI-Native Desk

2.1 A native desktop terminal

A cross-platform, branded desktop application — not a browser tab — that provides a Bloomberg-class cockpit for analysts and PMs. Real-time market data, charting, news, screening, position view, and agent interaction in one window.

2.2 A multi-agent fleet

Specialized agents for the work that humans currently do:

• Director / Strategy — frames the question, decomposes into tasks.

• Research — pulls filings, earnings calls, sell-side notes.

• Quant — technical, statistical, factor, and backtest analysis.

• Macro — interest rates, GDP, credit, monetary policy.

• On-chain / Sentiment — for crypto and retail-driven names.

• Risk — sizing, exposure, drawdown, scenario.

• Execution — order routing, paper trading, broker integration.

• Persona panel — bull / bear / Buffett / Graham / Lynch personas for adversarial review.

2.3 100+ real data connectors

Equities, crypto, futures, FX, macro (FRED, IMF, World Bank, DBnomics), alternatives (maritime, satellite, geopolitical), corporate filings, earnings call transcripts, on-chain whale flows, multi-venue crypto market data via CCXT.

2.4 Broker connectivity

Real and paper trading via Zerodha, Angel One, Alpaca, IBKR, plus 12+ additional venues. Strategy export to TradingView Pine v6, MT5, and TDX for hybrid workflows.

2.5 Persistent memory and bias detection

Every analyst’s decisions, theses, and outcomes feed a persistent trade-journal layer. The platform identifies systematic biases (over-trading, anchoring, loss aversion patterns) and surfaces them in weekly reviews.

2.6 Audit-grade logging

Every agent decision, every data pull, every order — recorded with inputs, intermediate reasoning, outputs, and timestamps. Replayable. Regulator-ready.

3. Reference Architecture

┌──────────────────────────────────────────────────────────────┐
│            Native Desktop Terminal (Customer Branded)        │
│                                                              │
│   Charting · News · Screener · Positions · Agent Workspace   │
└──────────────────┬───────────────────────────────────────────┘
                   │
        ┌──────────┴──────────────────────────────────────┐
        │            Agent Orchestrator                   │
        └──────────┬──────────┬──────────┬──────────┬─────┘
                   │          │          │          │
            ┌──────▼─┐ ┌──────▼─┐ ┌──────▼─┐ ┌──────▼─┐
            │Research│ │ Quant  │ │  Risk  │ │  Exec  │
            └──────┬─┘ └──────┬─┘ └──────┬─┘ └──────┬─┘
                   │          │          │          │
        ┌──────────▼──────────▼──────────▼──────────▼────┐
        │            Tooling & Data Layer                │
        │                                                │
        │  100+ Connectors · QuantLib · Backtests · MCP  │
        └──────────────────────┬─────────────────────────┘
                               │
        ┌──────────────────────▼─────────────────────────┐
        │     Governance, Risk & Compliance Layer        │
        │  Pre-trade checks · Kill-switch · Audit log    │
        │  RBAC · Four-eyes · Model risk mgmt (SR 11-7)  │
        └──────────────────────┬─────────────────────────┘
                               │
                  ┌────────────▼────────────┐
                  │   Broker / OMS / PMS    │
                  └─────────────────────────┘

Deployed inside the customer’s cloud, on-prem, or hybrid. Identity through customer SSO. Secrets in customer secret manager. AI inference can be routed to a customer-controlled model endpoint.

4. The Wohlig Productionization Layer

Open-source agent frameworks ship without the controls a regulated desk requires. Wohlig adds:

Model Risk Management — Aligned to SR 11-7, RBI/SEBI expectations. Documented model inventory, validation, monitoring.

Pre-trade risk and kill-switches — Hard limits on size, sector, exposure, drawdown. Automated halt on anomaly.

Best-execution audit — Order-routing rationale captured for every fill.

PII and data-residency controls — All data and inference inside customer perimeter.

Role-based access and four-eyes approval — High-risk theses and orders require dual sign-off.

LLM output guardrails — Respect SEBI IA / RA boundaries; suppress unauthorized advice patterns.

Immutable audit trails — Append-only logs for every agent decision and order.

Evaluation harness — Quality scoring on a held-out test set before any prompt or model change.

This layer is non-negotiable for any deployment that touches real capital, and it is where Wohlig’s experience makes the difference.

5. Outcomes Customers Can Expect

• 3–5x research throughput per analyst.

• 30–60% reduction in vendor data and terminal spend through feed consolidation.

• Idea-to-execution time compressed from days to hours.

• Improved risk discipline — every order pre-checked against exposure rules.

• Audit and regulator readiness out of the box.

• Bias-aware analyst development — measurable improvement in decision quality over time.

6. Engagement Model

Phase A — Foundation (6–8 weeks). Stand up the desktop terminal, data connectors, agent orchestrator, and governance layer in a paper-trading environment. Train two analysts.

Phase B — Pilot (8–12 weeks). Run real research with the platform alongside the existing stack. Validate quality, audit trails, and risk controls. Begin small live-money cohort if appropriate.

Phase C — Scale (ongoing). Roll out to the full investment team. Expand asset coverage, agent specialization, and broker connectivity. Wohlig stays as managed service or advisor as the customer prefers.

7. Target Customers

• Asset managers and AIFs (PMS, Cat-III hedge funds) in India and SEA.

• Family offices and UHNI desks.

• Crypto-native funds and prop desks.

• Fintech brokers and wealth platforms.

• Bank treasury, private banking, and wealth management.

• Sell-side research and equity-research boutiques.

8. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016, with offices in India and London. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Web: www.wohlig.com.

To discuss your investment-desk modernization, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Enterprise observability is structurally broken. Legacy APM platforms (Datadog, New Relic, AppDynamics, Dynatrace, Splunk APM) extract six-figure-per-year licensing fees while delivering symptoms rather than diagnoses. Engineers receive alert storms, not explanations. Data governance teams flag the export of telemetry to multi-tenant SaaS as a compliance risk.

A new generation of self-hosted, eBPF-native, AI-driven observability platforms changes the economics and the engineering reality. Wohlig builds and operates these platforms for customers across BFSI, healthcare, public sector, e-commerce, and SaaS — combining zero-instrumentation telemetry capture, SLO-driven alerting, and AI-powered Root Cause Analysis to cut MTTR by ~80%, observability TCO by 60–80%, and cloud waste by 15–30%, all while keeping telemetry inside the customer’s own perimeter.

This paper outlines the technical architecture, operating model, and business outcomes.

1. The Problem with Legacy APM

1.1 Cost

Per-host and per-GB pricing scales linearly with infrastructure. Most mid-to-large enterprises now report observability as one of their top three cloud-adjacent line items, often exceeding ₹5–25 crore per year.

1.2 Insight

Dashboards display what is wrong but rarely why. Engineers must correlate metrics, logs, and traces by hand under incident pressure. Senior SREs become single points of failure.

1.3 Coverage

SDK-based instrumentation is partial, drifts over time, and is inconsistently maintained across teams. New services launch with minimal telemetry.

1.4 Alerting

Threshold-based alerting produces storms during real incidents. On-call engineers learn to triage by “which service is hardest to silence” — the opposite of what you want.

1.5 Sovereignty

Sending production telemetry — which contains customer identifiers, transaction patterns, internal infrastructure layout, and security signals — to a third-party multi-tenant SaaS is increasingly unacceptable to legal, compliance, and CISO teams.

2. The Modern Pattern: AI-Powered, Self-Hosted Observability

2.1 Zero-instrumentation telemetry via eBPF

Rather than asking developers to instrument every line of code, eBPF agents capture network and system calls at the kernel level. They observe HTTP, gRPC, Postgres, MySQL, Redis, Kafka, and MongoDB traffic as it crosses the wire — automatically discovering services, mapping dependencies, and producing RED metrics (rate, errors, duration) on every edge from the moment they are deployed.

2.2 SLO-driven alerting

Alerts are bound to user-visible Service Level Objectives, not to raw threshold tripwires. The system emits one consolidated alert per service when an SLO is at risk — eliminating the alert storm pattern.

2.3 AI Root Cause Analysis

When an alert fires, an AI engine walks the discovered service dependency graph, correlates anomalies across metrics, logs, traces, and continuous profiles, and produces a narrative explanation identifying the failing pod, query, deploy, or upstream dependency. In Wohlig deployments, 80%+ of incidents are auto-explained at the moment of detection.

2.4 Continuous profiling

Code-level profiles run continuously and pinpoint the exact line burning CPU or RAM. Capacity and performance regressions are caught at commit-and-deploy time, not after a customer complains.

2.5 Deployment tracking

Every release is automatically compared against the previous baseline: latency, errors, and cost. Bad deploys are flagged within minutes, giving engineering an automatic rollback signal.

2.6 Data sovereignty

The entire stack — agents, storage, analysis, UI, AI engine — runs inside the customer’s own VPC or on-prem environment. Telemetry never leaves the perimeter. AI inference can be routed to a customer-controlled model endpoint for high-sensitivity environments.

3. Reference Architecture

┌───────────────────────────────────────────────────────────-┐
│                     Kubernetes Cluster(s)                  │
│                                                            │
│   [eBPF Agent]   [eBPF Agent]   [eBPF Agent]   ...         │
│        │              │              │                     │
└────────┼──────────────┼──────────────┼─────────────────────┘
         ▼              ▼              ▼
   ┌──────────────────────────────────────────────────-┐
   │   Observability Platform (customer VPC)           │
   │                                                   │
   │   ┌─────────┐  ┌────────────┐  ┌───────────────┐  │
   │   │Metrics  │  │ Logs/Traces│  │  Continuous   │  │
   │   │(Prom)   │  │ (ClickHouse)│ │  Profiles     │  │
   │   └────┬────┘  └─────┬──────┘  └───────┬────────┘ │
   │        │             │                 │          │
   │        └─────────────┼─────────────────┘          │
   │                      ▼                            │
   │           ┌──────────────────────┐                │
   │           │ AI RCA + Inspections │                │
   │           └──────────┬───────────┘                │
   │                      ▼                            │
   │           ┌──────────────────────┐                │
   │           │  SLO Alert Manager   │──▶ Pager/Chat  │
   │           └──────────────────────┘                │
   │                      ▼                            │
   │           ┌──────────────────────┐                │
   │           │  Single-Pane UI      │                │
   │           └──────────────────────┘                │
   └───────────────────────────────────────────────────┘

Stack: eBPF agent (DaemonSet), Prometheus-compatible metrics store, ClickHouse for logs/traces/profiles, OpenTelemetry-compatible ingest, Helm-based Kubernetes deployment, Apache 2.0 open-source core. Coexists with existing Prometheus / OTel collectors.

4. Business Outcomes

MTTR reduction — 60–85%

Observability TCO reduction — 60–80%

Cloud-cost reclaim from continuous profiling — 15–30%

Coverage of services with telemetry — from 40–60% (SDK-based) to ~100% (eBPF)

Alert volume reduction — 70–90%

Time-to-onboard a new service — from days/weeks to minutes

5. Where AI Specifically Earns Its Keep

The AI layer is not a marketing label. It is doing four concrete jobs:

1. Anomaly correlation across telemetry types — connecting a latency spike to a slow query to a CPU profile hotspot to a deploy event, in one pass.

2. Service-graph traversal — walking the dependency graph to find the originating failure rather than the loudest one.

3. Narrative generation — turning a multi-dimensional anomaly into a sentence a tired on-call engineer can read and act on.

4. Capacity and cost reasoning — proposing right-sizing actions from continuous profile and request data, with estimated savings.

Each is independently audited by Wohlig’s evaluation harness during deployment, with escalation paths back to human review for high-consequence decisions.

6. Engagement Model

Phase A — Foundation (3–4 weeks). Deploy the platform in a single non-production cluster. Validate eBPF compatibility, ingest, and AI RCA on a representative workload. Train the SRE team.

Phase B — Expansion (6–10 weeks). Roll into production clusters. Configure SLOs for top business services. Wire alert routing to PagerDuty / Opsgenie / Slack. Begin parallel-running with the legacy APM.

Phase C — Decommission and Operate (ongoing). Cut over from the legacy APM on a schedule. Wohlig stays as managed service, hybrid, or advisor depending on customer preference.

7. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have completed 20+ cloud migrations, shipped 10+ generative-AI solutions in production, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss your observability transformation, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

The modern marketing department runs on five to nine disconnected SaaS tools, an external performance agency, and a quarterly PDF audit that arrives weeks after it is useful. The result is structurally predictable: 20–40% paid-media waste, slow reporting, fragmented customer data, and a privacy posture that fails any serious regulator review.

Wohlig’s AI Marketing Command Center is a private, deployed-in-your-cloud platform that consolidates CRM, lead pipeline, outreach, content calendar, performance analytics, and a continuous AI-driven paid-media audit engine into a single console. It replaces a stack of subscriptions with a single governed platform owned by the customer, deployable on-prem for regulated industries.

This paper outlines the problem, the architecture, the audit methodology, and the engagement model.

1. The Problem with Today’s Marketing Stack

Wohlig’s discovery work across mid-market and enterprise marketing teams has identified five recurring pain points:

1. Tool sprawl. A typical marketing operation runs CRM (Salesforce / HubSpot), an email sequencer, a content calendar tool, paid-media reporting (Supermetrics / Looker), an audit tool, and increasingly an “AI assistant.” None of them share data cleanly.

2. Stale audits. Paid-media audits are quarterly, manual, and produced by external agencies. Issues persist for weeks before being discovered.

3. Spend leakage. Misconfigured pixels, broken CAPI / Events API integrations, untracked landing pages, audience overlap, and dayparting misfires routinely waste 20–40% of paid budget.

4. Privacy and sovereignty exposure. Sending lead lists, conversion data, and CRM exports through third-party SaaS conflicts with internal privacy policies and emerging regulation (DPDPA, GDPR, sectoral rules).

5. No board-ready reporting. CMOs spend the last week of every quarter reconciling spreadsheets to produce a single performance view for the board.

2. The Solution: One Private, AI-Driven Console

The Command Center brings four capability layers into one deployment:

2.1 Marketing Operations Layer

• CRM and lead pipeline management.

• Multi-channel outreach orchestration with audit logs and pause controls.

• Editorial and content calendar.

• Cron-style scheduled marketing automations (lead refresh, scoring, outreach windows).

2.2 Performance Analytics Layer

• Cross-platform spend, conversion, and revenue tracking.

• Attribution windows and customer-journey views.

• ROAS, CPA, LTV:CAC modeling at campaign and cohort granularity.

2.3 AI Audit Engine

• 250+ checks per platform across Google Ads, Meta, LinkedIn, TikTok, Microsoft Bing, Apple Search Ads, and YouTube.

• Industry-tuned templates (SaaS, e-commerce, B2B, healthcare, finance).

• 0–100 health score and A–F grade per platform.

• Prioritized fix list with estimated impact.

• A/B test design recommendations.

• Competitor creative intelligence.

• Landing page conformance and Consent Mode V2 / CAPI / Events API validation.

• Auto-generated client-ready PDF reports.

2.4 Agent Runtime

• Internal agents handle routine ops: weekly export pulls, lead re-scoring, content draft generation, audit scheduling.

• Each agent is specified, reviewed, and versioned (see Wohlig’s Governed AI Skill Factory whitepaper for the underlying methodology).

3. The Audit Methodology

The audit engine is the differentiator. It runs the following passes on each platform:

Tracking integrity — Pixel firing, CAPI / Events API health, server-side tagging, Consent Mode V2 compliance.

Account hygiene — Naming conventions, structure, negative keywords, audience overlaps, ad scheduling.

Creative quality — Ad-copy variants, image specs, CTR vs benchmarks, fatigue indicators.

Bid and budget — Bid strategy fit, budget pacing, dayparting, attribution window.

Landing experience — Page speed, mobile fitness, message match, form friction.

Financial model — CPA, ROAS, LTV:CAC modeling against industry benchmarks.

Competitor view — Public ad library scan, creative themes, frequency, positioning.

The output is normalized into a single 0–100 score per platform, a letter grade, and a prioritized list of actions sized by estimated business impact. Reports are generated as branded PDFs, dashboard widgets, and CRM-aware tasks.

4. Reference Architecture

                  ┌─────────────────────────────────────┐
                  │  AI Marketing Command Center (UI)   │
                  └──────────────┬──────────────────────┘
                                 │
          ┌──────────────────────┼──────────────────────┐
          │                      │                      │
   ┌──────▼──────┐       ┌───────▼──────┐       ┌──────▼─────-─┐
   │   CRM /     │       │  Performance │       │   Audit      │
   │   Pipeline  │       │  Analytics   │       │   Engine     │
   └─────────────┘       └──────┬───────┘       └──────┬───────┘
                                │                      │
                          ┌─────▼──────────────────────▼─────┐
                          │      Agent Runtime + Scheduler   │
                          └─────┬────────────────────────────┘
                                │
                       ┌────────┼────────┐
                       │        │        │
              ┌────────▼┐ ┌─────▼──┐ ┌──▼─────────┐
              │ Ad APIs │ │ CRM /  │ │ Email /    │
              │ (Live)  │ │ DWH    │ │ Channel    │
              └─────────┘ └────────┘ └────────────┘

Deployed inside the customer’s chosen cloud (GCP, AWS) or on-prem. Identity through customer SSO. Secrets in customer secret manager. Telemetry into customer observability stack. Data does not leave the customer’s environment.

5. Outcomes Wohlig Customers Can Expect

• Recover 15–30% of paid budget within the first two quarters by acting on continuous audit findings.

• Compress reporting cycles from quarterly to weekly without additional headcount.

• Replace 4–6 SaaS subscriptions with one platform — typically ROI-positive within 6–9 months on subscription savings alone.

• Pass internal privacy review because all data remains inside the customer’s tenancy.

• Give the CMO a board-ready scorecard generated automatically.

6. Engagement Model

Phase A — Discovery (2 weeks). Audit existing stack and spend profile. Identify the highest-leakage platforms and the highest-priority data integrations.

Phase B — Deploy (4–6 weeks). Stand up the command center in the customer’s cloud. Connect ad platforms, CRM, and email. Configure industry-specific audit templates.

Phase C — Optimize (ongoing). Run weekly audits, act on findings, expand to additional channels and brands. Wohlig stays as much or as little as the customer wants.

7. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com.

To discuss your marketing transformation roadmap, reach Wohlig at chintan@wohlig.com.

A Wohlig Transformations Whitepaper

Executive Summary

Enterprises are now past the question of whether to adopt generative AI. The new question is how to scale it safely. Most organizations get the first two or three AI use cases into production, then stall — held back by prompt sprawl, missing review processes, no versioning, and a growing risk surface that the CISO and CFO refuse to absorb.

Wohlig’s Governed AI Skill Factory is a platform pattern and engineering practice that turns the production of AI capabilities (we call them skills) into a repeatable, audited, multi-phase pipeline. It allows non-developers to specify new AI workflows safely, mass-produces those workflows with consistent quality, and gives the organization the audit, governance, and reuse it needs to operate AI at scale.

This paper describes the architecture, the engineering pipeline, the governance model, and the business outcomes Wohlig customers can expect.

1. The Problem: Why Enterprise AI Stalls After the First Wins

Wohlig has consistently observed five failure modes in enterprises that have crossed the AI proof-of-concept line:

1. Prompt sprawl. Dozens or hundreds of prompts live in scripts, notebooks, and Slack messages with no inventory or owner.

2. No review. Prompts go from a developer’s laptop to production with no formal quality gate. Hallucinations, prompt injections, and policy violations reach customers.

3. No versioning. A prompt that worked last quarter is silently degraded by a model upgrade. There is no way to roll back.

4. No reuse. Each team rebuilds the same RAG pattern, the same parser, the same summarizer — three to five times over.

5. No audit trail. When regulators or internal audit ask “what did this AI do, and on whose authority?” — the answer is a shrug.

The cost is real: failed pilots, blocked rollouts, abandoned investments, and growing executive scepticism after the initial enthusiasm.

2. The Solution Pattern: A Skill Factory

Wohlig’s Skill Factory is built on four principles:

• Specifications, not prompts. Every AI capability is described in a structured spec (inputs, outputs, tools, guardrails, examples) before any code is written.

• Pipelines, not heroics. Skills are produced through a deterministic multi-phase pipeline that any team can run.

• Review by default. No skill ships without passing automated and multi-agent reviews.

• Versioning end-to-end. Skills, prompts, evaluations, and rollouts are tracked the same way code and infrastructure are.

3. The Four-Phase Pipeline

Phase 1 — Intake and Triage

The factory accepts a natural-language request from any authorized user. A triage layer:

• Classifies intent (new skill, extension, replacement).

• Checks the existing skill catalog for overlap.

• Captures required inputs, outputs, data scopes, and risk class.

This eliminates the “another team already built this” duplication problem on day one.

Phase 2 — Multi-Lens Design

Every candidate skill is run through a structured design review using eleven reasoning lenses, including:

• First principles — what is the actual underlying need?

• Inversion — what would make this skill obviously bad?

• Pre-mortem — assume this fails in production. Why did it fail?

• Systems thinking — what upstream/downstream systems are affected?

• Adversarial — how would a malicious user abuse this?

The output is a hardened design document, not a prompt.

Phase 3 — Specification and Generation

The hardened design becomes a formal specification (machine-readable, e.g. XML or structured Markdown with frontmatter). The skill artifact — instructions, references, scripts, examples — is generated from the spec. Because generation is deterministic and spec-driven, a single source of truth governs both behavior and documentation.

Phase 4 — Multi-Agent Review

Specialized review agents check the generated skill in parallel:

• Design reviewer — does the skill match the spec?

• Usability reviewer — will the target user actually succeed?

• Evolution reviewer — is this maintainable?

• Script reviewer — are any tools or scripts safe?

A skill ships only when all reviewers approve. Failures are routed back to the appropriate phase with structured feedback.

4. Governance Model

The factory enforces governance the same way modern software platforms enforce CI/CD: through pipeline gates that cannot be bypassed.

Spec validation — Every skill must have a complete, parsable spec.

Policy review — PII handling, data egress, prompt-injection defenses.

Eval threshold — Skill must beat a quality bar on a held-out test set.

Cost ceiling — Per-call and per-month spend caps.

Multi-agent sign-off — All review agents must approve.

Human approval (high-risk) — Compliance officer countersign for risk-class 3+.

Versioned rollout — Canary, shadow, or blue-green deployment.

This is the governance layer that allows AI to enter regulated environments — BFSI, healthcare, public sector — without slowing innovation.

5. Reference Architecture

[ Intake UI ] ──▶ [ Triage Service ] ──▶ [ Skill Catalog Lookup ]
                                              │
                                              ▼
                                  [ Multi-Lens Design Engine ]
                                              │
                                              ▼
                                  [ Spec Repository (Git) ]
                                              │
                                              ▼
                                  [ Generation Service ] ──▶ [ Skill Artifact ]
                                              │
                                              ▼
                              [ Review Agent Pool (parallel) ]
                                              │
                                              ▼
                                  [ Evaluation Harness ]
                                              │
                                              ▼
                                  [ Release Service ]
                                  ├─ canary / shadow / prod
                                  └─ telemetry + cost guardrails

Wohlig deploys this on the customer’s chosen cloud (GCP, AWS) inside their own VPC. Identity flows through the customer’s existing SSO. Secrets sit in the customer’s secret manager. Telemetry flows into the customer’s existing observability stack.

6. Outcomes

In Wohlig engagements, customers consistently see:

• 70–90% reduction in time-to-ship a new AI workflow once the factory is operational.

• Order-of-magnitude reduction in unreviewed prompts in production.

• Documented audit trail for every AI capability — passes internal audit and regulator inquiry.

• Reusable skill catalog that compounds in value across departments.

• CFO-friendly cost discipline — every skill has a per-call ceiling and a monthly budget.

7. Engagement Model

Wohlig delivers the Skill Factory in three phases.

Phase A — Foundation (4–6 weeks). Stand up the factory infrastructure, spec format, and first three pilot skills. Train the customer’s AI CoE.

Phase B — Expansion (8–12 weeks). Migrate existing prompt assets into the factory. Onboard 3–5 departments. Establish the governance committee.

Phase C — Operate (ongoing). Wohlig stays involved at the level the customer wants — fully managed, hybrid, or pure advisory. Customers retain full ownership of skills, specs, and infrastructure.

8. About Wohlig

Wohlig Transformations is a digital transformation, cloud, and AI consulting firm founded in 2016. We have shipped 10+ generative-AI solutions in production, completed 20+ cloud migrations, and hold 40+ Google Cloud certifications including a Data Analytics specialization. We serve governments (Maharashtra, Gujarat, ONDC), enterprises (Lodha, Eros Now, Hungama), and high-growth consumer companies (Swiggy, Ninjacart, PW Live).

Offices: India and London. Web: www.wohlig.com

To discuss your AI scaling roadmap, reach Wohlig at chintan@wohlig.com.

Wohlig Transformations · AI Engineering

There is a quiet failure mode in enterprise AI that nobody puts on
the slide deck. You build an agent. It works. It saves time. The
team is delighted. You build five more. Three months later, half of
them are silently broken — an upstream API changed a field, a tool
schema drifted, an edge case nobody anticipated has crept into
production. The team is back to doing the work manually. Token
spend is up. Confidence in AI is down. The CFO is asking pointed
questions.

This is the agent decay problem, and it is the single biggest
reason enterprise AI initiatives lose momentum after the first wave
of wins.

The other reason — even more expensive — is token-bill creep.
Every agent invocation re-derives the same reasoning the agent did
yesterday. The model thinks through the tax filing logic from
scratch. Then thinks through the compliance check from scratch. Then
thinks through the contract clause from scratch. Multiply by ten
thousand invocations a month, and you are paying for the same
thinking, repeatedly, in perpetuity.

Wohlig builds the fix. We call it a Self-Improving Agent
Platform.

The pattern

Three properties make agents compound rather than decay:

1. Skill capture. When an agent successfully completes a task,
the workflow is captured as a named, versioned, reusable skill.
The next time a similar task arrives, the agent does not re-reason
from scratch — it retrieves the skill and runs it. This alone cuts
token spend by 30–50%, with the savings compounding as the skill
library grows.

2. Self-repair. Skills are monitored continuously. When a skill
fails — because a tool changed, a schema drifted, or an edge case
appeared — the platform diagnoses the failure and patches the skill
automatically. No engineer chasing the regression. No quiet decay.
The agent is back online while the team is still at lunch.

3. Shared registry. Skills are stored in a governed registry
that every agent in the organization can read from. When the tax
team’s agent learns to handle a new GST schema, the audit team’s
agent and the legal team’s agent benefit immediately. One team’s
work compounds into capability for everyone else.

The result is what AI was always supposed to deliver: capability
that grows with use, instead of decaying.

What changes for the business

Token spend collapses. Industry data on this pattern shows
roughly 46% reduction in token usage. For an enterprise running
serious agent workloads, that is real money — and the savings grow
with adoption rather than shrinking.

Reliability stops being a fire drill. Agents that used to
silently break and trigger emergency engineering investigations now
self-heal. The engineering team focuses on building new capability
rather than maintaining last year’s.

Knowledge stops being trapped. The clever agent the marketing
team built that nobody else knows about now lives in the registry,
discoverable, reusable, and auditable. Capability accumulates across
the organization.

ROI becomes measurable. Every skill carries execution history,
version lineage, and cost-per-run metrics. When the CFO asks “what
did we save with AI this quarter?” — the answer is a real number
with a permalink.

One agent, many surfaces. A skill written once works whether
the agent is invoked from WhatsApp, Slack, Microsoft Teams, an
internal portal, a CLI, or a desktop integration. The platform
abstracts away the surface — your customers and employees use it
where they already are.

Where this works

This is not for a team running their first AI pilot. This is for
organizations that have already shipped multiple AI workflows and
are now staring at:

• A token bill that has climbed every month for six months.

• A backlog of “agent X is broken again” tickets.

• A growing realization that the same capability is being rebuilt
  three times in three departments.

If that sounds familiar — particularly in BFSI, professional
services, BPO/KPO, or document-heavy enterprises — the
self-improving pattern is the next architectural step.

Where Wohlig fits

We do four things:

1. Build the platform. The skill engine, the registry, the
   self-repair loop, the multi-surface gateway, the dashboard.
   Deployed inside your cloud, governed by your identity, audited
   by you.

2. Migrate your existing agents. Whatever you have built so far
   — copilots, chatbots, automation scripts — gets wrapped as
   skills, versioned, and brought into the registry.

3. Train your AI team. Engineers and domain experts get the
   patterns and the playbooks for authoring durable skills.

4. Operate it with you. As much or as little as you want —
   managed service, hybrid, or pure advisor.

The honest summary

The first wave of enterprise AI was about getting agents into
production. The second wave — the one that actually delivers ROI —
is about making those agents compound rather than decay. Skill
reuse cuts cost. Self-repair preserves reliability. A shared
registry turns one team’s wins into the whole organization’s
capability.

Wohlig builds this. If your AI program is at the stage where the
costs are obvious and the returns are getting harder to point to,
this is what comes next.

Wohlig Transformations builds AI, cloud, and data platforms for
governments, enterprises, and high-growth startups. 10+ generative-AI
solutions in production. Founded 2016. Offices in India and London.

Wohlig Transformations · Product Engineering

For the last five years, the answer to “we need to edit video” has
been: pay $20–$50 per seat per month to one of a dozen browser-based
SaaS editors. Veed. Kapwing. Descript. Canva Video. The footage gets
uploaded to their cloud, edited there, rendered there, and downloaded
back. The seats multiply. The bill grows. And every BFSI, healthcare,
and legal customer who tries to use any of them runs into the same
brick wall: we cannot send raw client footage to a third-party
multi-tenant cloud.

Meanwhile, every D2C brand and ed-tech platform with serious
content velocity has the opposite problem — they want to embed a
video editor inside their own product so creators do not bounce out
to a separate tool. Building one from scratch is a 12–18 month
engineering exercise that nobody can spare the team for.

Wohlig builds the way out of both problems. We call it a Private
Video Editor.

What it is

A browser-based, multi-track video editor that runs entirely on the
client side. Users open a URL. They drag video, audio, images, and
text onto a multi-track timeline. They trim, split, duplicate, layer.
They preview in real time. They export to MP4 up to 1080p. The entire
encoding pipeline runs in the browser through WebAssembly. No upload
to a third-party cloud. No rendering backend.

That last property is the one that closes deals. The footage never
leaves the user’s machine — or, in the deployed-internally version,
never leaves the customer’s network. There is nothing to leak. There
is no third-party data-processing agreement to negotiate. The CISO
signs off in one meeting.

Wohlig deploys this in two patterns:

Pattern A — Embedded inside your product. A SaaS, ed-tech, or
creator-economy platform gets a fully white-labeled editor running
inside its own UI. Branded. Themed. Wired into the product’s auth and
storage. A capability the product team would otherwise spend
12–18 months building.

Pattern B — Private corporate editor. A regulated enterprise
gets a self-hosted editor, deployed in the customer’s VPC or on-prem,
gated by SSO. Marketing, internal comms, sales enablement, training
— all the corners of the business that produce video — get a tool
they can actually use without uploading content to an external SaaS.

Where it wins

1. Cost. A 100-seat creator team at ₹2,500/seat/month is paying
~₹30 lakh per year in SaaS video subscriptions. The private editor
costs the deployment infrastructure plus a one-time engineering
engagement.

2. Sovereignty. Footage never leaves the customer’s perimeter.
For BFSI client testimonial videos, healthcare patient content,
legal recordings, and government communications, this is the
unblocker.

3. Velocity. Marketing teams stop waiting for an agency to turn
around a 30-second cut. Sales teams record and trim demo clips
themselves. Ed-tech instructors edit lecture footage between
classes. Internal comms teams cut town-hall recaps in an hour.

4. White-label revenue line. SaaS companies in the creator,
ed-tech, podcast, and short-video adjacent categories add an
embedded editor as a feature — without building it from scratch.
Differentiates the product, eliminates the bounce-out to an external
tool, increases time-in-product.

5. Extensibility. Once the foundation is in place, Wohlig can
layer AI features on top: auto-cut on silence for podcasts,
transcript-driven editing for interviews, auto-captioning, brand-
consistent intro/outro stitching, generative B-roll, batch export
for ad creatives at every aspect ratio.

Where this fits

• D2C and e-commerce brands producing high-volume product reels,
  UGC edits, and ad creatives — and bleeding ROAS waiting on
  external production cycles.

• Creator-economy and short-form-video SaaS wanting an embedded
  editor as a feature module.

• Ed-tech and L&D platforms where instructors trim lectures,
  add captions, and assemble lesson videos in-browser.

• Internal corporate comms / HR / IC teams at large enterprises
  producing town halls, onboarding, and training clips.

• Digital agencies and MarTech vendors wanting a white-label
  editor inside client portals.

• Media, news, and sports orgs that need fast clip-and-publish
  workflows.

What Wohlig actually delivers

The browser-based foundation is open-source and client-side. What
Wohlig brings is the production engineering that turns it into a
real platform:

• Custom branding, theming, and product integration.

• Auth, identity, storage, and DAM/CMS hooks.

• Server-side render farm for 4K, long-form, and batch jobs that
  exceed browser limits.

• AI features — auto-captioning, silence detection, transcript-
  driven cutting, generative B-roll, brand asset enforcement.

• Aspect-ratio presets for Reels, Shorts, YouTube, OTT, and
  programmatic display.

• One-click export pipelines into the customer’s content stack.

• VPC / on-prem deployment for regulated environments.

• Performance tuning for long timelines and large media libraries.

The honest summary

The video tooling industry has spent five years quietly billing
enterprises for the privilege of uploading their own footage to
someone else’s cloud. The technology to run a serious editor
entirely in the browser, or entirely inside a private network, is
now mature enough to flip that equation.

If you are paying per-seat for video SaaS — or if you are about to
build an editor inside your own product — Wohlig has the
foundation, the engineering, and the deployment patterns to do it
better, faster, and on infrastructure you actually own.

Wohlig Transformations builds AI, cloud, and data platforms for
governments, enterprises, and high-growth startups. 10+ generative-AI
solutions in production. 40+ Google Cloud certifications. Founded
2016. Offices in India and London.

Wohlig Transformations · Data Engineering

The dashboard sprawl in the average enterprise data team is, at this
point, a running joke. Eighty active dashboards, half of them
unused. Three different definitions of “active customer.” Two
parallel revenue numbers that disagree by 4%. A finance team that has
quietly given up and gone back to a spreadsheet. And a Tableau or
Looker bill that climbs every quarter as more “viewer seats” get
provisioned for people who only check one number once a week.

The core problem is not the tool. It is the model. Drag-and-drop BI
suites were built in the 2000s for a world where the analyst was the
bottleneck. In 2026, the bottleneck is governance — too many
definitions, too many dashboards, no single source of truth, and no
real audit trail when a number changes.

Wohlig’s recommendation for any company that has crossed this line:
stop buying BI seats and start treating analytics as code.

What “BI in Git” actually means

Imagine your KPIs not as drag-and-drop widgets in a vendor SaaS, but
as Markdown files in your own Git repository. Each file describes a
report — copy, charts, filters — using plain SQL queries against
your warehouse. To change a metric, you open a pull request. To roll
back a wrong definition, you git revert. To audit the history of
“net revenue”, you read the commit log.

When the file is built, it produces a fast static website that
business users open in a browser. Pages load in milliseconds because
the heavy lifting happened at build time. Read seats are unlimited
because there are no read seats — it is a website. The analytical
engine pushes down to your existing warehouse (Snowflake, BigQuery,
Redshift, Databricks, Postgres) and uses an embedded columnar query
layer for sub-second performance on the front-end.

Wohlig deploys this pattern inside the customer’s own cloud,
connected to the customer’s own warehouse, with the customer’s own
SSO and identity. Nothing leaves the perimeter.

What changes for the business

1. Licensing cost collapses. A mid-size enterprise running 200
Tableau viewer seats at ~$15/seat/month is paying ~₹30L per year for
people to look at numbers. The code-first pattern costs whatever
your warehouse and CDN cost — and adds zero per-seat licensing.

2. One source of truth, enforceable. Every metric is defined
once, in version-controlled SQL, reviewed via pull request. The
data team stops being the referee in revenue-definition arguments —
the answer is a permalink to the file.

3. Analytics ships at engineering speed. New reports in hours,
not weeks. No BI-admin ticket queue. Analysts who already write SQL
and use Git become productive on day one.

4. Audit and governance become free side effects. Git history is
the audit trail. Pull requests are the change-management workflow.
SOC 2, RBI, HIPAA, and DPDP teams stop asking “who approved this
change to the customer-acquisition-cost calculation” because the
answer is in the merge commit.

5. Embedded analytics for free. Want branded dashboards inside
your own SaaS product? It is a static-site embed. No second BI tool,
no second contract, no separate access model.

6. Vendor lock-in vanishes. Your reports are markdown and SQL.
They are portable, diffable, copyable. Your data stays in your
warehouse. There is no proprietary file format to escape.

Where this works

• Mid-market and enterprise data teams already on a modern
  cloud warehouse (Snowflake, BigQuery, Redshift, Databricks) and
  comfortable with SQL and Git.

• Engineering-led organizations — fintech, SaaS, e-commerce,
  logistics — where the analytics function reports into engineering.

• Companies actively trying to consolidate or replace Tableau,
  Looker, or Power BI on cost grounds.

• Product teams wanting embedded white-labelled dashboards
  inside their own SaaS — without spinning up a separate BI vendor.

• Regulated industries (BFSI, healthcare) needing on-prem or
  VPC-hosted BI with full audit trail and no per-seat licensing.

What Wohlig adds

The code-first pattern is genuine open-source — Wohlig does not
charge license fees for it, and we say that openly. What Wohlig
charges for is the engineering that turns it into an enterprise-
grade analytics platform:

• Warehouse architecture, source connectors, and query optimization.

• A governed metric layer — semantic definitions, reusable models,
  test coverage on critical KPIs.

• CI/CD for analytics: lint, dry-run, peer review, deploy on merge.

• SSO, role-based access, row-level security where required.

• Embedded analytics inside your product — branded, themed,
  per-tenant.

• Migration from your existing Tableau / Looker / Power BI estate —
  including consolidating duplicate definitions on the way over.

• An AI-chat-over-data layer for non-technical executives sitting on
  top of the same governed SQL models.

We have been delivering this stack for customers across e-commerce,
ed-tech, BFSI, and government. We will deliver it for you in your
cloud, with your team, on your warehouse — and leave you fully
self-sufficient.

The honest summary

Drag-and-drop BI was the right answer in 2010. In 2026, the same
companies that have moved their code, their infrastructure, and
their security posture into Git are still letting their analytics
live in a vendor SaaS with no diff, no review, no rollback, and a
per-seat bill that goes up every year.

The fix is structural. Move BI into the same engineering workflow as
the rest of your business. Wohlig builds it.

Wohlig Transformations builds AI, cloud, and data platforms for
governments, enterprises, and high-growth startups. 40+ Google Cloud
certifications, including a Data Analytics specialization. 20+ cloud
migrations delivered. Founded 2016. Offices in India and London.