As companies continue to figure out how to adapt software development for the AI era, the software factory model—an agent loop built around traditional development stages—has gained traction. Warp is now stepping in with a new offering designed to simplify that transition. On Tuesday, the AI coding company introduced Warp Factories, an infrastructure layer that provides a ready-made environment for deploying agents and a roadmap for using them.

The system comes pre-configured with standard development phases—triage, specification, implementation, review, and verification—and allows users to automate any of these steps. It supports different coding models, including Codex and Claude Code, and integrates with tools like Linear, Jira, Slack, and Teams to fit into existing workflows.

Warp CEO Zach Lloyd says the target market is smaller companies that lack the resources to build such a system from scratch. He noted that running agents in the cloud, steering them, setting up shared memory, and creating evals is a significant infrastructure undertaking. Warp Factories handles that heavy lifting out of the box.

The system also gives managers visibility into performance metrics and token spend across agents, and it includes self-improvement loops to optimize the process. While Warp Factories can automate a substantial portion of development work, Lloyd emphasized it is not meant to replace engineers. He said Warp currently automates about 30-35% of its tasks on a weekly basis, and he expects that number to rise as models and tools improve.

Notably, some companies like Stripe and Ramp have already built similar internal systems with success, but Warp's offering aims to make the model accessible to those without such in-house capabilities.

Source: TechCrunch