Notes & essays
What changed when Shopify merchants could describe an objective through Claude instead of translating it into dashboard settings. A reflection on MCP, configuration fatigue, and product onboarding from my work with Dynamatic.
2026-10-05 · 11 min readBuilding a dashboard for the questions a person actually asks about an agent system: current health, usage, sessions, judgment, daily briefings, and the record of actions. Includes the gap between an action running and its outcome being verified.
2026-08-23 · 5 min readHow designing for software agents changes assumptions about interfaces, contracts, and context. A reflection on the difference between information a machine can parse and information a person can understand.
2026-08-18 · 8 min readLiving with an agent system through provider failures, fallback behavior, persistent context, and maintenance. A practical account of Hermes and the Alfred configuration, including the limits of what was verified.
2026-08-17 · 16 min readA structured approach to stateless AI agents using explicit context, state tracking, and decision loops. Explores why repeated actions and lost context need system design rather than another prompt.
2025-08-12 · 4 min readA personal account of using NotebookLM mind maps and audio discussions while studying. Covers organizing source material, making difficult concepts approachable, and checking AI-generated explanations.
2025-07-30 · 7 min read