Blog

Notes on building software that works

Insights on AI agents, engineering, and getting real value from AI.

  1. Aug 12, 2026 1 min read

    Your Agent Fleet Doesn't Need a Better Model — It Needs an Org Chart

    The bottleneck in agentic systems is authority and decision rights, not model capability — borrowed from leadership doctrine

  2. Aug 12, 2026 1 min read

    How to Read an AI Provider's Data Terms in 15 Minutes

    A practical, five-question method for reading a provider's data terms — the tier line does most of the arguing

  3. Aug 12, 2026 1 min read

    Greenfield Is the New Default — How AI Rewrote the Build-vs-Rebuild Decision

    AI collapsed the cost of writing code, so the honest default flips to greenfield — with the five conditions where brownfield still wins

  4. Aug 12, 2026 1 min read

    The Five Rungs of AI Adoption: Why the Bottleneck Is Never the Tool

    Adoption is near-universal and depth is rare — the constraint at every rung is human, not technical

  5. Aug 12, 2026 1 min read

    Decide Determinism First — The Question That Shapes Every AI Agent

    Determinism is a premium you pay per task, not per agent — decide it first and the architecture falls out of it

  6. Aug 12, 2026 1 min read

    Copilot Studio vs. Claude: Which AI Platform Makes Sense at Your Company Size

    Company size and Microsoft maturity shape the right answer more than features do

  7. Aug 12, 2026 1 min read

    Microsoft Copilot Studio vs. Claude Enterprise: What the Numbers Actually Say

    Direct capability and cost comparison — condensed analyst brief

  8. Aug 12, 2026 1 min read

    The Best AI Tool for Your Business Is the One Your Team Will Actually Use

    AI adoption fails at the people layer, not the technology layer