Building with AI

I lead technology teams by day, but I stay close to the craft. These are tools and systems I have built hands-on with Claude — to keep my understanding of AI first-hand, not second-hand.

• A personal AI working environment

  A customized Claude environment I use as a daily thinking and work partner. It remembers context across sessions, follows my preferences and decision-making frameworks, and routes complex work to specialized helpers — turning a general-purpose AI into a reliable collaborator.

• An LLM-powered knowledge base

  A system that ingests documents and notes and lets me ask questions of them in plain language, with Notion as the store. It turns scattered material into something I can actually interrogate and reason with.

• Versioned memory & skills backup

  The discipline layer beneath the environment above: its accumulated memory and custom skills are version-controlled and backed up, so that knowledge is preserved and portable rather than trapped on one machine.

• A marathon training agent

  An assistant that plans and adjusts my endurance training from wearable data, connecting each week’s load to recovery and longevity science. Notion holds the data; Claude does the planning.

• AI playtesting for games (playcheck)

  An experiment in using AI to judge whether a game is actually fun, not just functional — a different kind of quality test that looks at the experience rather than correctness.

• Automated video creation workflow for creators

  An end-to-end pipeline that automates short-form video production — from idea and script to a finished, publish-ready clip — so a creator can go from concept to output without the manual production overhead.

• Content research engine

  A tool that researches what makes content resonate — surfacing the angles, framings, and openings most likely to land with an audience — so writing starts from evidence rather than a blank page.