About Dr. Harish Kotadia, Ph.D.

Enterprise Agentic AI Architect · Author

Off the clock. Caricature by request — the enterprise architecture is the day job.

I design governed, production-grade agentic AI systems for regulated industries. I help CIOs and CTOs move agentic AI from pilot to production. Systems that survive a board review, an auditor, and a regulator. Not slideware. Systems that run.

I have spent 20+ years building enterprise technology for Fortune 10 and Fortune 500 clients. I did not pivot to AI. I arrived through every wave that made it possible:

CRM and ERP at scale
Data warehousing and business intelligence
Big data and predictive analytics
Cloud AI, then generative AI
Agentic systems on Claude and MCP today

That history means I read an agentic AI decision the way you have to: against a real budget, real risk, and a board that wants the business case.

In July 2026 I published Intent In, Outcomes Out: A Practitioner’s Field Guide to Agentic AI for the Enterprise. It is the argument this site makes, written down in one place.

What a CIO or CTO gets

Four things. They are the four most programs are missing when they stall.

Production, not proof of concept. Systems built for regulated environments. Human-in-the-loop review and audit-grade observability from day one.
A business case the board can read. My doctoral training was in measuring business value. So I ask the question engineers skip: what is the return, and how is it measured.
Governance as a first principle. Data sovereignty, regulatory constraints, and accountability designed in. Not bolted on after an incident.
Full-stack architecture depth. Multi-agent orchestration, tool integration, agent identity, execution boundaries, observability.

Proof in production

Two systems define what I mean. Both run in regulated industries at enterprise scale.

Regulated power utility — on-premises generative AI platform

Meta’s Llama under strict data-sovereignty and NERC compliance constraints
RAG-based field technician copilot
Regulatory correspondence assistant
Llama · RAG · on-premises · NERC compliance

Major auto-loan originator — end-to-end agentic system

Claude as the core reasoning engine
Multi-agent orchestration through MCP
Human-in-the-loop review consoles
Audit-grade observability, designed in from the start
Claude · MCP · human-in-the-loop · observability

What I publish

Everything is free except the book.

Book — Intent In, Outcomes Out. The practitioner’s field guide. Kindle and paperback, July 2026.
Case study library — 507 verified enterprise deployments. Searchable by industry, platform, and roadmap stage.
Newsletter — The Agentic AI Architect. One control per edition, proven with a paired failure and success.
Newsletter — Enterprise Agentic AI Case Studies. One documented deployment per edition. What it cost, what it returned.
Frameworks — The definition and the Five-Stage Roadmap, Prompted to Autonomous.

Recognition

Thinkers360 named me a top thought leader in five categories for 2026:

Top 10 — AI Orchestration
Top 10 — AI Safety
Top 25 — Agentic AI
Top 50 — AI Infrastructure
Top 100 — AI Governance

Earlier: KDnuggets Top 13 Big Data influencers. Published on social CRM and customer analytics in CustomerThink and Social Media Today.

The path here

I led 200-person delivery teams on CRM and ERP programs for Fortune 10 financial services clients in the early 2000s. I built data foundations several of those institutions still run on. Every wave since, hands on.

My Ph.D. in Marketing Management was doctoral research in marketing analytics. Regression modeling and statistical analysis of client satisfaction, long before “data science” was a job title. It is why I ask what a system returns before I ask what it runs on.

Who this site is for

I write for the people who have to decide: CIOs, CTOs, enterprise architects, consultants, and product leaders. Most writing on agentic AI is too academic or too shallow to act on. Every post here is grounded in what was actually built and what held up when the stakes were high.

Start with the definition. Then the roadmap. Then the case studies. The argument builds from definition to consequence to action.

Work with me

I am open to senior architecture, advisory, and consulting engagements with enterprises and IT services firms deploying agentic AI in regulated and high-stakes environments. The fastest way to reach me is a LinkedIn message.

LinkedIn → X / @AgenticAIArch → Amazon Author →

Key West, ninety miles from Cuba. The only place my inbox has ever lost an argument.

Southernmost point, Key West.