---
date: "2026-09-26T13:40:25.000Z"
title: "2026-09-26"
location: "Brooklyn"
draft: false
---

It seems both important and valuable to measure how long it takes an agent to make a change in a codebase and how this changes over time

---

Can we create integration and end to end test from logs (and by instrumenting logs) and from metrics to create a foundational environment on top of which we can reliably evolve a codebase (with agents) without breaking it