How I think about technology that has to pay off.

Data, AI and LLMs, architecture and cloud costs, seen from the business: what they cost, what they return, and when they are worth it.

Sep 05, 2026·11 min read

Incorrect Data Is Dangerous. Ambiguous Data Is Too.

A metric can be perfectly calculated and still lead to a bad decision. Ambiguous definitions, time differences, and misinterpreted metrics can cause different teams to believe they're talking about the same thing when they aren't. And as AI agents begin to act on that data, precision stops being just an analytical concern: it becomes an operational one.

also inESPT