Factory

Factory is an agent-native software development platform centered on “Droids,” its software development agents. The fetched homepage says “Your software Factory powered by Droid” and points developers to a CLI install command, docs, enterprise, security, and open-source resources. The pricing page is more detailed than the initial staging file: Pro is $20/month, Plus is $100/month, Max is $200/month, and Teams is custom for growing teams.

The individual tiers explain Factory’s product shape. Pro includes a desktop, CLI, SDK, cloud and local background agents, billing and usage statistics, and an agent-readiness dashboard. Plus adds expanded rolling rate limits, about 5x the usage of Pro, and Droid Computers, Factory cloud computers for remote Droids. Max adds about 10x Pro usage and early access to new features. Teams includes multiple members up to 150 seats, custom usage limits, dedicated onboarding and support, SSO, and SAML/SCIM provisioning. That is a serious developer-tool posture rather than a generic AI assistant.

Teams should evaluate Factory that want agents to own bounded engineering tasks, not just autocomplete. It competes with Cursor Agent, Devin, Aider, and GitHub Copilot Agents. The differentiator is the “Droid” workflow across desktop, CLI, SDK, local/background execution, and managed cloud computers. That makes it interesting for teams experimenting with asynchronous agent work: issue cleanup, migration assistance, test generation, refactors, or implementation of well-scoped tickets.

The honest risk is operational control. Background coding agents can create a lot of change fast, and fast code is not by default good code. Start with a repository that has strong tests, narrow permissions, clear coding standards, and mandatory human review. Track accepted versus rejected patches, time saved, and bug regressions. Heavy users should also model usage tier jumps: $20/month is easy to approve, but $100 or $200/month per individual needs a clear productivity case.

Frame a Factory pilot around asynchronous work. Give a Droid a ticket with clear acceptance criteria, a test command, and a small blast radius; then compare its pull request against a human baseline. The platform’s value is highest if background agents can finish useful slices while engineers focus on design and review. It is lower if developers spend more time correcting broad, underspecified changes. Teams should also use the agent-readiness dashboard and billing statistics from Pro before moving into Plus or Max. Those metrics help separate real engineering payoff from novelty-driven usage.