Thoughts
Thoughts
Thoughts on context, AI and building in public. Short, no fluff. written for myself, public by default.
- The difference between an AI assistant and an autonomous agentThoughts · 2026-11-19
- Why agent payments need an append-only audit trailThoughts · 2026-11-16
- Least privilege for AI agentsThoughts · 2026-11-12
- What an agent authorization record should containThoughts · 2026-11-09
- The credit card is the wrong interface for AI agentsThoughts · 2026-11-05
- Deterministic authorization for AI agentsThoughts · 2026-11-02
- What it means to give an AI agent a budgetThoughts · 2026-10-29
- An AI agent should never have your credit cardThoughts · 2026-10-26
- What should an AI agent be allowed to doThoughts · 2026-10-22
- The authorization layer for autonomous agentsThoughts · 2026-10-19
- Why there should be no LLM in the authorization pathThoughts · 2026-10-15
- Who is responsible when an AI agent buys the wrong thingThoughts · 2026-10-12
- Why I'm Building Know ComputerThoughts · 2026-10-09
- Why AI agents keep stopping at checkoutThoughts · 2026-10-05
- Why AI agents need permission systemsThoughts · 2026-10-05
- Welcoming Raj Koli, Founding Engineer (Agent Experience): the interviewThoughts · 2026-10-01
- Welcoming Alex Mwaniki, Founding Engineer (Core): the interviewThoughts · 2026-09-22
- Finding Ghassen: the co-founder question, answered in three weeksThoughts · 2026-09-21
- Artifact-based hiring: ship code before titlesThoughts · 2026-09-20
- The money layer for AI agentsThoughts · 2026-09-20