How Warp ships 2,000 PRs a month with AI factories | Zach Lloyd (CEO, Warp)

The "software factory" represents a paradigm shift in development, transitioning from siloed local coding to a centralized, agent-driven lifecycle. By integrating AI agents across the entire workflow—from initial task kickoff in Slack to automated issue tracking, code generation, and computer-use verification—engineering teams can achieve greater velocity and transparency. Zach Lloyd, CEO of Warp, demonstrates how this approach allows managers to measure performance through metrics like human interactions per pull request, identifying bottlenecks where human intervention slows progress. Beyond engineering, these factory-style workflows extend to non-technical tasks, such as automating slide deck design via Figma or analyzing sales call transcripts to refine product positioning. This model emphasizes data-driven optimization, where LLM-based judges evaluate agent performance, enabling continuous improvement of the factory's configuration and reducing operational costs.