Harness Engineering: How to Build a Software Factory — Dru Knox, Tessl
- Source
- youtube.com
- Author
- AI Engineer
- Date
Context
A talk at the AI Engineer World's Fair 2026 proposes ways to tell whether a team has gone from using coding agents to running a software factory, one where agents write everything users get and engineers build the factory. In the talk, Dru Knox, head of product and design at Tessl, defines a factory by three measures: autonomy, automation and quality, and argues the real payoff is higher quality, not just speed.
Knox calls the discipline of getting there engineering, which some now call loop engineering, and splits the work into inner, outer and meta loops. The metrics the talk lists to track are manual takeovers by a human, human comments on pull requests, and pull requests started without a human. Tessl sells tools aimed at this work, and the published summary reports no measured results, so the talk is best read as a proposed way of keeping score.
- Knox defines a software factory by three metrics, autonomy, automation and quality, and argues the real payoff is higher quality rather than just speed.
- The talk splits harness engineering into inner, outer and meta loops, and three layers to build: a control plane, an -ready stack and improvement loops.
- Suggested signals for factory maturity: manual takeovers, human PR comments, and PRs started without a human.
- Knox says the work is hard because the field changes weekly, the work is unplanned, and the signals teams need are stuck in local logs.
- agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…






