#roi · 4 essays
AI ROI — Baselines Before Benefits
Most AI value claims fail a simple question: measured against what? If nobody wrote down the before, the after is a story. These essays are about the discipline that makes AI value undeniable — baselines before benefits, unit economics before scale, and budgets that don't tax the engineers who found the value.
Start with AI ROI — If You Didn't Baseline It, It Didn't Happen, then Rationing Tokens Is Not an AI Cost Strategy for the budget side. The pilot triage makes unit economics one of its five dimensions — a pilot whose cost per task doesn't survive volume gets the verdict it deserves.
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.
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 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.
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.