Factory drops Droids to write production code on its own

Writing code is usually the easy part of software engineering. The real headache is deciphering ticket threads, chasing Sentry errors, and matching requirements across Notion docs.

Factory has officially opened general access to Droids, which it calls the world's first autonomous software development agents.

Wiring AI into the entire dev stack

Factory argues that software development goes far beyond generating code snippets. Instead of acting as another inline autocomplete tool, Droids are built to autonomously produce production-ready software.

To give them the full picture, Factory connects directly into a company's broader engineering stack, pulling context from GitHub, Slack, Linear, Notion, and Sentry. That means the agent can theoretically understand what broke in production, read the relevant team discussion, and write a targeted fix.

Factory also claims to be the first platform supporting both local synchronous workflows and remote asynchronous agent execution, allowing developers to either pair with Droids live or offload tasks in the background.

Why it matters

AI coding assistants are clearly shifting from passive autocomplete extensions into active, background workers.

We think Factory's approach is smart because the hardest context to capture in engineering has always lived outside the codebase, scattered across crash trackers and issue boards. Wiring that entire toolchain into an agent moves AI development closer to how human engineers actually work.

That said, autonomous production-ready code raises obvious reliability questions. Letting an asynchronous agent push changes based on scattered Slack threads sounds convenient, but teams will likely want strict guardrails and human reviews before trusting them anywhere near production systems.