Essays by Burak Galiba
Notes from the field
On getting enterprise AI out of pilot purgatory and into the P&L. No slideware.
Start here
New to the site? The opening essay first, then the agent trilogy in order.
Or start from a topic page: #ai-strategy · #leadership · #governance · #pilots · #transformation · #roi · #agentic-ai
You Can't Write a Five-Year AI Plan. Write a Five-Year Commitment.
AI capability turns over in months while strategy plans run five years. The fix isn't a shorter plan — it's sorting every line by how long it stays true, and committing only to the half that survives.
Whose Number Is It? Why IT Should Never Own the AI Value Line
When the AI benefit line in the strategy plan belongs to IT, adoption stalls and finance can't audit the claim. The value is only real when business units own the number.
The AI Capacity Fork — More Throughput or a Different Team Shape
If AI creates capacity there are two honest destinations for it, and most organisations pick neither. On reconfiguring the unit of delivery, and cutting elastic capacity in the right order.
Rationing Tokens Is Not an AI Cost Strategy
Uber burned its 2026 AI budget in four months, then capped spend per engineer. The rationing cycle is predictable, and a uniform cap taxes exactly the people who found the value.
Your Documents Are Not AI-Ready — and Retrieval Will Not Fix That
Twenty years of documentation is inventory, not a corpus. Why retrieval blends contradictions a human would catch — and the ten-question test that proves it.
Your Business Unit Became an AI Pilot Factory — How to Govern It Back Into Production
Monthly workshops, showcase meetings, thirty agents that are really system prompts, nothing live. How a business unit governs its own AI work into production.
Applause Is Not an Operating Model — How AI Demos Get Declared "Ready" Before the Real Work Starts
The demo was flawless because the data was frozen. How AI demos get declared "ready" at exec meetings — and the intake process that would stop it.
Change Management Is the Project, Not the Postscript
McKinsey tested 25 attributes; workflow redesign moved EBIT most. Why the announcement-email rollout fails, and what shipping the change actually looks like.
AI in Name Only — How Enterprises Pick Flagship AI Projects That Aren't AI
Enterprise AI portfolios fail before day one — at selection, where every incentive rewards relabeling analytics as AI. Here's the test, and the fix.
Your AI Agents Are Stalled at IT — The Access Layer Nobody Built
Your tier model is published, but agents have nothing to connect to. Why direct database access fails, and how a thin, governed access layer fixes it.
The Hackathon-to-Production Gap — A Paved Road for the AI Agents Your Business Units Already Built
Business units built AI agents at a hackathon. Now they're stalled at IT. Why the fix is a paved road: tiered access, published SLAs, and real dates.
Kill More AI Pilots. Seriously.
Gartner says 40% of agentic AI projects will be canceled by 2027. Good. Killing pilots early and on purpose is how winners fund what ships.
AI Governance Stopped Being Optional — TRAIGA, Colorado ADMT, and the EU Clock
Texas TRAIGA is in force, Colorado's ADMT law lands January 2027, EU dates are set. Why NIST AI RMF is the safe-harbor move that survives it all.
Do You Need a Chief AI Officer?
76% of organizations now have a Chief AI Officer, up from 26% a year ago. When the role earns full-time headcount, when fractional wins, how to decide.
AI ROI — If You Didn't Baseline It, It Didn't Happen
95% of AI pilots can't prove ROI, mostly because nobody measured anything first. A simple baseline discipline that makes AI value undeniable.
Everyone Has AI Agents. Almost Nobody Has Them in Production.
Most enterprises have adopted AI agents. Few run one in production — and Gartner expects 40% of agentic projects canceled by 2027. Mind the gap.
Your AI Advisor Is Selling You Their Stack
Model vendors run consulting arms, clouds embed free engineers, the Big 4 rent you agents. Everyone advising enterprises now has inventory to move.
Pilot Purgatory — Why Your AI Demos Never Ship, and How to Escape
Your AI demos impress everyone and ship nothing. Why pilots stall between demo and production — and the triage that gets one live in 90 days.
What Is AI Transformation, Really — and What Do Enterprises Actually Get?
AI transformation is not a tool rollout. A working definition, the four layers that make it real, and how to tell the genuine article from theater.
5 Mistakes Enterprises Make With AI — the Same Five, Year After Year
Five mistakes that show up in enterprise AI portfolios again and again — unowned pilots, vendor-written strategy, governance procrastination — and the fix for each.
Why 95% of AI Pilots Fail — and What the 5% Do Differently
An MIT NANDA report found 95% of enterprise GenAI pilots deliver zero P&L impact. The failure is organizational, not technical — here's what the other 5% do.