Blog
Where I share what I'm thinking about — technology leadership, AI, data, and the craft of building software. Written to be useful, not to fill a calendar.
If you have to bark a direct order, the failure usually happened earlier. Mission, plan ownership, decision training, and real support are what make decentralized command work—for teams and for agent systems.
Read moreAsk an AI to pick a number from 1 to 30 and 17 appears suspiciously often. That small quirk reveals a much bigger lesson about how language models work.
Read moreA successful AI demo proves possibility. Production requires a bounded workflow with ownership, evaluation, authority, observability, and a fallback.
Read moreSchool taught many of us that mistakes cost points. Effective teams treat bounded mistakes as information, then turn that information into better decisions and systems.
Read moreGraph engineering gives AI a trustworthy map of business entities, relationships, provenance, and time. A practical guide to building the context layer beyond vector search.
Read moreCode review is not only a bug-catching gate. It is how a team understands, accepts, and retains ownership of every change entering its codebase.
Read moreBefore an AI agent touches real work, define its owner, outcomes, authority, access, evaluation, and offboarding plan.
Read moreManual cloud changes feel fast until you need repeatability, accountability, or compliance evidence. Build the smallest paved road early.
Read moreThe honest signals that say it's time to bring in executive technology leadership — and the moments when it's premature.
Read more