Dr. Harish Kotadia Ph.D.

10 Facts You Wish You Knew Before Building Agentic AI Solution
Linkedin
August 29, 2026

This article was originally published here: https://agenticaiarch.com/agentic-ai-facts/ It is republished here with permission. Read the original for the latest updates.

AI-Native SDLC Playbook Staged on Agentic AI Roadmap
Linkedin
August 29, 2026

This article was originally published on agenticaiarch.com: https://agenticaiarch.com/ai-native-sdlc-roadmap/. It is republished here with permission. Read the original for the latest updates.

Managing the Harness Across the Five-Stage Roadmap
Linkedin
August 28, 2026

This blog post was originally published here: https://agenticaiarch.com/agentic-ai-governance-harness/

Agentic AI Rewires IT Services Firms
Linkedin
August 22, 2026

Originally published on my blog site here: https://agenticaiarch.com/agentic-ai-it-services-shakeout/

The CxO power shakeout over agentic AI will force IT services and consulting companies to rebuild their sales, marketing, and delivery model around business executives — CFOs, COOs, CMOs — instead of the CIO and CTO organizations they have sold to for thirty years, and to staff those accounts with deep domain experts rather than technology professionals. That is the argument of this post, and I will make it bluntly: the firms that keep running the CIO playbook will watch their accounts get taken by firms that learned to sit in the CFO’s office and talk about cost per completed task.

IT Services Pricing: From Billable Hours to Billable Decisions
Linkedin
August 13, 2026

This article was originally published on my site: https://agenticaiarch.com/price-per-decision-agentic-ai-it-services-pricing/

A price-per-decision model charges for a completed unit of agentic work — a resolved case, a closed loop — instead of the hours it took a person to get there or the seats a company happened to buy. That’s the short version. The longer version: this isn’t really a pricing change. It’s an admission that the thing being sold has changed, and most of the industry hasn’t said so out loud yet.

Why I Wrote Agentic AI Book Earned Autonomy
Linkedin
August 01, 2026

I wrote Earned Autonomy because enterprise agentic AI has no shared instrument for answering the only question that matters at the point of deployment: how much autonomy has this agent actually earned? The book supplies one. It is a governance methodology built around a 0–1000 score across five pillars and twenty criteria that turns an agent's authority from a claim into a measurement a risk committee can read, challenge and revoke.

Agentic AI Just Undid 30 Years of IT Discipline
Linkedin
July 26, 2026

Originally published by Dr. Harish Kotadia, Ph.D. on AgenticAIArch.com: https://agenticaiarch.com/agentic-ai-just-undid-30-years-of-it-discipline/

Vibe coding is having its moment. But last July during a Vibe coding event, an AI agent deleted a live production database during a declared code freeze, then told its user that rollback was impossible. It wasn’t.

That is not a model failure. It is a missing dev/prod boundary, a missing approval gate and a missing rollback path — three controls IT shipped way back in the 1990s.

Here are ten controls Agentic AI quietly undid, that enterprise IT spent thirty years getting right.

What Is Agentic AI Technology? A Practitioner’s Field Map of the Enterprise Stack
Import from wordpress feed
July 19, 2026

“Agentic AI technology is the software stack that lets an enterprise hand a system an outcome to achieve rather than a script to execute”  © Dr. Harish Kotadia, 2026 — the model, memory, tools, identity, guardrails, and observability that turn a goal into governed, auditable action.

Token Troubles: Five Enterprises That Learned the Hard Way
Linkedln
July 17, 2026

That is the paradox defining agentic AI economics in 2026, and it is not a pricing problem. It is an architecture problem. A chatbot answers a question with one inference call. An agent decomposes a task, selects tools, calls sub-agents, validates outputs, retries on failure — and every one of those steps burns tokens. A leading industry analyst firm's March 2026 analysis found agentic workflows consume 5 to 30 times more tokens per task than a standard chatbot query. Some research puts the multiplier far higher.

Why Agentic AI Budgets Are Surging in 2026 — and the Fix
Import from wordpress feed
July 13, 2026

Agentic AI budgets are breaking in 2026, and not because the models are bad. They are breaking because enterprises are deploying agents without mapping token spend to projects and outcomes. Agentic AI cost attribution is the practice of tagging every token an agent consumes to a named project, workf

Why I Wrote This Book on Agentic AI
Import from wordpress feed
July 12, 2026

Over the past eight months, I have had the opportunity to speak with senior technology professionals at Fortune 500 companies while evangelizing agentic AI solutions. What I realized surprised me: many of them were not doing agentic AI right. Not because of lack of talent or budget, but because they

Two Agentic AI Deployments, One Contained, One Catastrophic
Import from wordpress feed
July 07, 2026

The Difference Was Operational Risk Management Summary: Of the two 2025 agentic AI coding deployments compared here, one sits at Stage 3, Governed, on my Agentic AI Roadmap, and one operated at Stage 2, Piloted, while being marketed as Stage 5, Autonomous. The gap between them was not model capabili

11 OpenAI Agentic AI Case Studies Analyzed and Mapped
Import from wordpress feed
July 06, 2026

Quick answer Of the 11 OpenAI case studies analyzed here: 1 sits at Prompted, 3 at Piloted, 3 at Governed, 3 at Assured, and 1 at Autonomous. This is the most evenly distributed maturity spread across OpenAI, AWS, and Claude vendor case studies analyzed in this series. Companies that reach Assured o

Agentic AI Autonomy Trap: Healthcare vs Fintech Case Study
Import from wordpress feed
July 05, 2026

Two Agentic AI Case Studies – One Success, One Reversal And What Separates Them I have spent this year documenting agentic AI wins. This edition is different. I want to put a genuine success story next to a genuine reversal, side by side, because the gap between them is not about the technolog

Agentic AI’s Autonomy Trap – A Case Study
Import from wordpress feed
July 05, 2026

I have spent this year writing about agentic AI wins across banking, healthcare, and manufacturing. A European fintech company is the case I keep coming back to for a different reason. It is the clearest public record of a company that pushed straight past Governed toward a claimed Autonomous, skipp

Agentic AI in Healthcare: The Prior Authorization Case Study
Import from wordpress feed
July 04, 2026

For as long as I have been building agentic AI systems, I have heard a version of the same objection when the conversation turns to healthcare. Protected health information cannot touch an AI agent, they say. The liability is too high, the compliance burden too strict, the margin for a confident wro

Agentic AI in Regulated Industries: Case Studies Proving Skeptics Wrong
Import from wordpress feed
July 03, 2026

The ABA Banking Journal recently reported on the broader shift now underway across financial services, from Anthropic’s ten ready-to-run agent templates for pitchbooks, KYC screening, and month-end close, to a major wealth management firm’s advisor note-taking agent, to another wealth ma

The Agentic AI Roadmap Stage Behind Tesla Robotaxi Headline
Import from wordpress feed
July 03, 2026

Tesla turned on unsupervised Robotaxi service in Miami. No safety monitor in the seat, live in a geofenced zone covering West Miami, Doral, and Coral Gables. Florida is now the first state outside Texas to get it, and Miami is the fourth city in the network after Austin, Dallas, and Houston. I read

From Siebel to Agentic AI: How Enterprise Data Finally Got Free
Linkedln
June 29, 2026

I have spent over twenty years installing software so that business people could look at their own customer data. I mean that literally. Some of my earliest projects involved putting a CRM client on individual desktops, one machine at a time, before a salesperson could see a single account they already owned. That should sound strange to you. It took me a long time to find it strange too.

11 AWS Agentic AI Case Studies Analyzed and Mapped
Linkedln
June 28, 2026

I spent this weekend going through the AWS Solutions Library and its customer success stories, hunting for the deployments that have actually shipped. I was specifically interested in AWS agentic AI case studies—not the polished demos, but the ones running in production with real numbers attached to them.

Agentic AI Assurance Is Now a Market, for REAL
Linkedln
June 26, 2026

Something new is showing up in enterprise AI budgets. Not agents. The audit of agents.

A market is forming around agent assurance. Independent third-party review. Runtime monitoring. Even guardian agents whose only job is to watch other agents and rein them in when they drift.

The Agentic AI Roadmap: 5 Levels From Prompted to Autonomous
Linkedln
June 26, 2026

Instructions in, results out. That was IT. Intent in, outcomes out. That is agentic AI. But an organization does not make that turn all at once. It matures into it. Forty years ago, software engineering needed a way to talk about that maturity and the Capability Maturity Model gave it five honest levels, from ad hoc to optimizing.

Who Was That? The Agent Identity Question Nobody Can Answer
Linkedln
June 25, 2026

For most enterprises, there is no good answer. In a 2026 survey the Cloud Security Alliance ran with Aembit, more than two-thirds of organizations could not clearly distinguish an action taken by an AI agent from one taken by a human. Sixty-eight percent. The agent does the work, and the log records a person. That is not a reporting gap. It is an accountability gap, and accountability is the whole game in a regulated workflow.

Agentic AI Reality Check: The 40% That Will Fail
Linkedln
June 24, 2026

The loudest number in agentic AI right now is a warning. Gartner says more than 40 percent of agentic AI projects will be canceled by the end of 2027. The reasons are not exotic. Escalating costs. Unclear business value. Risk controls that were never built. The same research calls out "agent washing," and estimates that only about 130 of the thousands of vendors claiming to sell agents actually do.

Agentic AI Case Study: Auto Loan Processing on Claude
Linkedln
June 18, 2026

I have spent close to two decades architecting enterprise systems for Fortune 100 clients. The auto loan platform I built around Anthropic Claude was different. The AI did not sit on the side as a feature. It sat in the middle as the operating model. Here is an honest account of how it came together, what it delivered, where it bit me and what I would change.

Govern the Loop, Not the Developer.
Linkedin
September 04, 2026

Agentic AI Governance — Edition 8
This edition is based on my new blog post, Agentic AI SDLC vs Traditional SDLC: What Changes. It opens with the definition of agentic AI SDLC:

Intent Describes. The Hook Enforces.
Linkedin
September 03, 2026

"A hook makes it mandatory, or it does not exist."

That line is from my new blog post, What is Agentic AI Hook? Control an Agent Cannot Talk Its Way Past. This edition is the short version.

Agile in Agentic AI SDLC
Linkedin
September 02, 2026

Agentic AI Governance — Edition 7
This edition is based on my blog post, Agile in Agentic SDLC? Keep the Ceremony. Retire the Artifact.

Design What Cannot Happen
Linkedin
August 31, 2026

Agentic AI Governance — Edition 6
This edition is based on my new blog post, What Is Agentic AI Architecture? The Six Layers That Actually Decide Outcomes.

Rent the Model, Own the Bill
Linkedin
August 30, 2026

Agentic AI P&L — Edition 8
Agentic AI, as I define it, is a governed, goal-driven software layer in which LLM-powered agents, equipped with memory, tools, and orchestration, plan and execute multi-step business processes on their own. Instructions in, results out was IT. Intent in, outcomes out is agentic AI. The full definition, built from a review of fifty published ones, is at What is Agentic AI?

Govern the Harness, Not the Model
Linkedin
August 28, 2026

Agentic AI Governance — Edition 5

This edition is based on my new blog post: https://agenticaiarch.com/agentic-ai-governance-harness/

Code Is No Longer the Bottleneck, Your Approval Gates Are.
Linkedin
August 28, 2026

This edition is based on my new blog post, Anthropic's AI-Native SDLC Playbook, Mapped to My Agentic AI Playbook.

The Governance Gap in Agentic AI
Linkedin
August 27, 2026

Agentic AI Governance — Edition 4
This edition is based on my new blog post: https://agenticaiarch.com/governance-gap-agentic-ai/

How Fortune 1000 Companies Organize for Agentic AI
Linkedln
August 26, 2026

The technology rarely kills enterprise agentic AI projects. The organization does. I have spent the past year building agentic AI systems for loan origination, and I keep seeing the same failure. So I went through the analyst research and the trade press to answer one question: how are large enterprises structuring themselves for agentic AI, and which setup is winning?

27 Key Agentic AI Terms: What They Mean
Linkedin
August 25, 2026

Agentic AI Architect — Edition 21
This edition is based on my new blog post: 27 Key Agentic AI Terms, What They Mean and Where They Came From.

The vocabulary of agentic AI has moved faster than any enterprise glossary I have seen in twenty years of consulting. Terms that were lab jargon eighteen months ago now sit in Gartner press releases, vendor pricing pages, and board decks. Often with three competing definitions attached.

Agentic AI Model Harness: Scaffolding Around the Model Decides Who Wins
Linkedin
August 24, 2026

Originally published on my blog: https://agenticaiarch.com/ai-model-harness/

What Is an AI Model Harness?

An AI model harness is everything wrapped around a raw language model that turns it into a working system — the system prompts, tool interfaces, context and memory management, the agent loop, orchestration, guardrails, permissions, sandboxing, and observability. The model weights are not the harness. And in 2026, the harness decides real-world outcomes at least as much as the model does.

Enterprise Agentic AI Case Studies — Edition 10
Linkedin
August 24, 2026

Six Months per Agent Became Four Days, Because One Review Now Covers Fifty
Bottom line: The North American arm of one of the world's largest automakers runs more than 50 AI agents in production on a single internal platform. Building a new agent used to take six engineers six months. It now takes one engineer four days. The design decision that explains it: security, architecture, and data review happen once, at the platform level, and every agent inherits the result. The operational numbers hold up under scrutiny. Intent in, outcomes out — and here the verified outcome is shipping velocity.

The C-Suite Power Shakeout Over Agentic AI — and What It Does to IT Services Firms
Linkedin
August 23, 2026

Agentic AI Governance — Edition 1
The CIO holds primary AI purchasing authority in 13% of organizations. 72% of CEOs now call themselves the main AI decision-maker — double last year. Those two numbers, side by side, are the whole story of this edition: the executives who fund agentic AI are about to become the executives who manage it, and I'm putting a date on it.

Agentic AI Shakeout Rewires IT Services Firms
Linkedin
August 23, 2026

This edition is based on my new blog post, How the CxO Shakeout Rewires IT Services Firms.

Yesterday I published my analysis of the coming CxO power shakeout — the transfer of agent fleet ownership from technology leadership to the business executives whose P&L the agents actually touch. That post was written for the enterprise side of the table. This one is for the vendor side. And the argument is blunt: firms that keep running the CIO playbook will watch their accounts get taken by firms that learned to sit in the CFO's office and talk about cost per completed task.

CxO Power Shakeout Over Agentic AI
Linkedin
August 22, 2026

Agentic AI P&L — Edition 7
This edition is based on my blog post: The Coming CxO Power Shakeout Over Agentic AI. Here is the argument in short.

Cheaper Tokens, Bigger Bills
Linkedin
August 21, 2026

Agentic AI Architect — Edition 20
In early 2026, a global ride-hailing company gave coding agents to its engineers. Adoption jumped from about a third of its 5,000 engineers to 84 percent in months.

Enterprise Agentic AI Case Studies — Edition 8
Linkedin
August 20, 2026

One Billion Interactions Later, the Only Cost Control That Scaled Sits in the Gateway
Bottom line: A top U.S. bank crossed one billion cumulative interactions on its agentic banking assistant in under three years, and the design decision that made that volume survivable is Control 2 of the six-control FinOps prescription I published this week: the model tier is decided in the orchestration layer, not in application code a developer can override.

The People Who Cap Agent Spend Don't Answer to the CFO
Linkedin
August 19, 2026

The number: 8% of FinOps practices report to the CFO. 78% report to the CTO or CIO — and 98% of them now manage AI spend, up from 63% a year ago.

Agentic AI P&L — Edition 5
Linkedln
August 18, 2026

The number: Over 40% of agentic AI projects will be canceled by the end of 2027, per Gartner's June 2025 forecast — for escalating costs, unclear business value or inadequate risk controls. All three should be addressed in the budget request, before a single token is bought.

The Biggest Model Should Never Be the Default
Linkedin
August 18, 2026

In May 2026, an engineer published a bill almost nobody believed. He had run a fleet of roughly 100 coding agents for one month. They consumed 603 billion tokens across 7.6 million requests.

Enterprise Agentic AI Case Studies — Edition 7
Linkedin
August 16, 2026

You are Responsible for what the Agent Says and Does

Enterprise Agentic AI Vision 2030: What Actually Changes
Linkedln
August 15, 2026

Enterprise agentic AI doesn't move in a straight line. It moves like a J-curve — two rough years of pilots and cancellations, then a scramble to scale, then, if the forecasts hold, an operating model that looks nothing like the IT services industry of 2025. Most analysts, academics, and IT services leaders agree on that shape. Where they split, often loudly, is how fast it moves and how far it actually gets.

The Governance Gap Is the Real Agentic AI Story
Linkedin
August 15, 2026

Gartner expects at least 15% of day-to-day work decisions to be made autonomously through agentic AI by 2028, up from effectively zero in 2024, and it expects agentic capability built into a third of enterprise software over the same stretch. I read that stat three or four times before it landed. Not because the number is shocking — it's the direction I'd assumed for two years — but because of what sits next to it. Deloitte surveyed 3,235 leaders across 24 countries and found that only 21% of organizations have a mature governance model for agentic AI. Three quarters expect to be running agents within two years. About one in five can actually say, with a straight face, that they know how those agents are supervised.

Enterprise Agentic AI Case Studies — Edition 6
Linkedln
August 14, 2026

A top-10 US bank runs a four-agent framework where a dedicated evaluator agent simulates every plan before it executes and rejects anything that violates policy — the design choice behind a reported 55% jump in dealer-site engagement and a 5x cut in response latency. Three months later, a US vertical SaaS platform's coding agent used an over-scoped API token to wipe its production database and every backup in nine seconds, because its safety rules lived only in a prompt the agent could read, acknowledge and override. Same technology category, opposite governance decision, opposite outcome. Intent in, outcomes out — the outcome is set by what you build to check the agent, not by what you ask it to do.

Per-Decision Cost: The Metric Your IT Budget Doesn't Have
Linkedin
August 13, 2026

This newsletter edition is based on my blog post originally published here: https://agenticaiarch.com/for-agentic-ai-compute-is-not-a-fixed-line-item/

Agentic AI does not cost what your IT budget assumes it costs.

Traditional IT financial management treats compute as a fixed, forecastable line item — CapEx depreciated over years or OpEx metered against a workload you defined in advance. Agentic AI breaks that assumption at the root: cost now scales with autonomy and reasoning depth, one decision at a time, driven by behavior your procurement team never modeled.

Earned Autonomy: A Governance Methodology and Scoring Framework for Agentic AI
Independently published (Amazon KDP)
July 30, 2026

Autonomy is not a setting you enable. It is a position your organization earns — and the evidence says almost nobody has earned it yet.

Gartner predicts that more than 40 percent of agentic AI projects will be canceled by the end of 2027, and it names the causes in a revealing order: escalating costs first, unclear business value second, inadequate risk controls third. Model capability is not on the list at all. Only 23 percent of organizations have scaled an agentic system anywhere in the enterprise. And two-thirds of the technology chiefs surveyed by the IBM Institute for Business Value report the same thing in different words: the authority they have handed to agents has outrun the accountability they can exercise.

That gap is what this book closes.

A stage-gated methodology, not a manifesto

Earned Autonomy sets out five stages — Prompted, Piloted, Governed, Assured, Autonomous — and assesses three tracks across them: governance, agent identity, and security and execution boundaries. Two meters run beneath every stage: what the work costs, and what it was worth. Each stage has a testable exit gate. Nothing advances on assertion.

An Index, and 507 deployments scored on it

The Earned Autonomy Index scores a deployment from 0 to 1000 across five pillars and twenty criteria, then applies one rule that changes the answer: your band is gated by your weakest pillar, not licensed by your highest total. Applied to 507 publicly documented enterprise deployments, three quarters sit in the Piloted band and not one reaches Autonomous. Two pillars — Outcome Assurance and Intent Specification — cap 470 of the 507. The field has built the controls that stop an agent doing damage. It has not built the controls that would tell anyone the agent has stopped doing its job.

Inside

The supervision gap: why authority outruns accountability, and what it costs
Agent identity: the control plane an examiner asks about first
The four-question control plane: what an agent may do, whose words it may obey, what it may hand on
The Token Trap: engineering agentic economics before the invoice arrives
Ten parameters for agent performance, and how each one gets gamed
The audit-survivability test: for any action any agent took, can you show who acted, on whose authority, and at what cost?
A self-assessment that gives you a score, a capping pillar, and a next move
Who it is for

CIOs, CTOs, enterprise architects, and risk and audit leaders who must decide how much autonomy to grant — and defend the decision afterwards.

I have spent more than two decades building enterprise systems for Fortune 100 clients, most recently the agentic AI platforms behind regulated auto loan and credit card origination. The methodology in this book is what that work taught me.

Companion volume to Intent In, Outcomes Out: A Practitioner's Field Guide to Agentic AI for the Enterprise.

Intent In, Outcomes Out: A Practitioner's Field Guide to Agentic AI for the Enterprise
Independently published (Amazon KDP)
July 10, 2026

A practitioner's field guide to building and governing agentic AI systems in the enterprise. Drawing on production experience in regulated auto loan and credit card origination, this book introduces the Five-Stage Agentic AI Roadmap and makes one argument: the model is a commodity — the loop, governed from day one, is the product.

Enterprise Agentic AI Case Studies
LinkedIn
July 24, 2026

I write Enterprise Agentic AI Case Studies, a LinkedIn newsletter that examines one publicly documented enterprise agentic AI deployment per edition. Each analysis covers the business problem, the agent architecture and stack, the governance and control model, the measured outcome, and what I would do differently. Sources are named and claims are traceable, drawn from my library of 507 verified enterprise agentic AI cases.

Agentic AI P&L
LinkedIn
July 10, 2026

Enterprise agentic AI, costed. Payback periods, hidden failure costs and the P&L math.