The software factory concept—an agent loop wrapped around the traditional stages of development—has become the early blueprint for companies rebuilding engineering orgs around AI. Stripe built its “minions” system in-house. Ramp deployed a background agent that watches its own code post-launch. Both required serious infrastructure work. Warp’s new product, Warp Factories, announced Tuesday, is aimed at everyone else.
Warp CEO Zach Lloyd is direct about the gap. “Things like running your agents in the cloud and steering those agents as they run, or bringing the work that they’re doing into your local environment, or setting up memory that goes across those agents, or setting up evals that go across those agents—it’s actually a huge infrastructure undertaking to do this right,” he told TechCrunch. Warp Factories ships with that architecture pre-built: triage, specification, implementation, review, and verification phases, any of which can be automated. It works with Codex or Claude Code, and plugs into Linear, Jira, Slack, and Teams.
What’s interesting here isn’t the novelty of the factory model—it’s the packaging. Warp is essentially productizing what Stripe and Ramp had to engineer from scratch, which matters for the long tail of companies that can’t afford a dedicated AI infrastructure team. The system also tracks token spend and agent performance metrics across configurations, closing a feedback loop that most ad-hoc setups lack.
Lloyd’s own numbers suggest realistic expectations. “We automate like 30% of our tasks, 30 to 35% on a weekly basis,” he said, predicting that figure will climb as models and context improve. That 30% figure is worth sitting with. It’s meaningful, but it’s also a long way from the full-autonomy narrative that often surrounds AI coding tools. Warp isn’t selling a replacement for engineers—just a cleaner way to manage the agents that are already showing up in pull requests.
