Journal

Long-form on AI engineering and venture studios.

Pieces we write when something is worth writing down properly. Mostly about how engineering organizations actually absorb AI, with the occasional note on how we run the studio.

  1. Studio thesis 9 min

    How we think about AI venture studios

    A venture studio is not a fund and not a lab. Here is the operating model Carmentara uses to build independent AI companies on shared infrastructure, and why we think this shape compounds better than alternatives.

    Read
  2. AI engineering 7 min

    The reorg that breaks in month two: a role boundary that keeps collapsing

    AI-era reorganizations fail at the same seam every time. It is a single role boundary that looks reasonable on the org chart and collapses in practice. By month two it is a retention problem.

    Read
  3. AI engineering 8 min

    The readiness mirage: why test coverage does not predict AI readiness

    Test coverage, documentation, and recency feel like they should predict how well a repository absorbs AI agents. They don't. Something else does, and almost no organization is tracking it.

    Read
  4. AI engineering 7 min

    The silent suppressor: the misconfiguration that caps every AI coding rollout

    There is a layer every engineering team ships into production before their code ever reaches an AI agent. It is misconfigured in roughly 7 teams out of 7. Nobody audits it because nobody sees it.

    Read