Where Factory has helped me most isn’t “write the feature from a one-liner,” but the slower, grittier work around the feature: tracing why something breaks, updating call sites after a signature change, fixing tests that fail for boring reasons, and drafting the mechanical parts of a change once the approach is already decided. I tend to use Droids in shorter loops. I’ll point one at a failing test, a stack trace, or a narrow refactor, then let it explore the repo, propose edits, and run checks while I stay on the decision-making side. That works better for me than long, unsupervised runs. It also matches how I already work in the terminal and GitHub, so it doesn’t force me into a new “home” for development. When the task is well-bounded and the success criteria are obvious, it reliably saves me an hour or two of tedious navigation and editing.

Factory AI is helpful for automating development work. It save my time, reduce manual tasks and help me complete coding work faster. The workflow is easy and useful for daily development.

I’ve been using Factory (Droids) for about six months on a backend-heavy product team at a mid-market SaaS company (around 25 engineers). The biggest difference for us is that, when a Droid is well scoped, it can take a ticket from start to finish instead of only suggesting the next few lines of code. The most common wins have been implementing clearly defined API endpoints or service changes, updating tests after a refactor, cleaning up modules that span multiple files, and producing a draft PR where the relevant context has already been explored. Because it can search the codebase, edit across files, run commands, and leave behind something reviewable, it cuts down on the number of small-to-medium tickets that would otherwise pull senior engineers into constant interruptions. I also appreciate that we can run it from the terminal or plug it into existing workflows, rather than being pushed into a single IDE. The autonomy controls have been genuinely helpful too: we keep tighter limits around production-sensitive areas, while giving it more freedom on internal tools and test code.