The Model Should Be Replaceable. The Agent Should Persist.
Why AI agents should persist even as the underlying models change.
Writing
Notes, walkthroughs, and things worth writing down from work in computer vision, robotics, and applied AI.
Why AI agents should persist even as the underlying models change.
When multiple AI agents collaborate, separate logs cannot explain the complete workflow. Effective observability requires a shared trace that follows each request across agents, tools, and handoffs.

A practical look at how AI, computer vision, and deterministic geometry can work together to turn messy real-world perception into reliable measurements a machine can actually use.