Blog

Articles & Notes

Product engineering, system boundaries, and the tools I ship with.

20

posts/2026

Latest

Recent writing tagged Agents.

Two Small Products, Two Hard Boundaries

Agentic UI and GitGrid shipped in the same week. Their useful features are different, but both products became clearer when I defined exactly what they are allowed to change.

Week of Aug 9, 2026published Aug 28, 20264 min read

Agent Approval Is a Ladder, Not a Switch

Cursor 3.6’s layered Auto-review design is a useful reminder that agent autonomy should depend on the action, the sandbox, and the available evidence—not one global yes-or-no setting.

Week of May 24, 2026published Aug 28, 20263 min read

Gemini 3.5 Flash and the Managed-Agent Shift

Google's Gemini 3.5 launch paired a new model with managed execution, showing how much of an agent product now lives outside the model itself.

Week of May 17, 2026published Aug 28, 20264 min read

Codex on Mobile Is an Approval Surface

Bringing Codex to a phone matters less for writing code than for steering, reviewing, and approving asynchronous work at the right moment.

Week of May 10, 2026published Aug 28, 20264 min read

Archive

2026

2 posts

When a Coding Agent Becomes Infrastructure

Cursor's SDK exposed its agent harness as a programmable service, making the infrastructure around the model the more interesting product boundary.

Week of May 3, 2026published Aug 28, 20264 min read

GPT-5.5 and the Long-Horizon Test

GPT-5.5 made long-running agent work the headline. The harder question is how to evaluate whether an agent can finish responsibly.

Week of Apr 26, 2026published Aug 28, 20264 min read

Get infrequent updates on new projects.