Factory was built around the thesis that AI coding tools should cover the full software engineering workflow, not just writing code. Its Droids are autonomous agents specialised for specific SDLC functions: a coding Droid implements features; a review Droid analyses PRs; an incident Droid triages on-call alerts and proposes fixes; a migration Droid modernises legacy code. All Droids share an organisational memory that indexes the codebase, documentation, Slack history, Jira tickets, and Notion pages.
Key capabilities
Full SDLC Droids — Specialised agents covering feature development, code review, security scanning, documentation generation, codebase modernisation, dependency updates, and incident response. Each Droid is tuned for its specific task domain rather than attempting to handle all tasks with a general-purpose agent.
Organisational memory — Droids index and continuously update a knowledge base spanning code repositories, engineering docs, Slack conversations, Jira/Linear tickets, and Notion pages — giving them the contextual understanding of a senior engineer who has been at the company for years.
Interface-agnostic deployment — Droids operate across CLI, VS Code and JetBrains IDE extensions, browser, Slack, Linear/Jira, and mobile. A Sentry alert can autonomously trigger a Droid that investigates the stack trace, proposes a fix, and opens a draft PR — all without a developer initiating the process.
Factory Bridge — A local execution bridge that lets Droids run in the developer's environment for tasks requiring local context, or hand off to Factory's cloud Droid Computers for background asynchronous execution.
Autonomy level
Level 5 — Fully autonomous. Droids accept a task description (from Jira, Slack, GitHub, or direct assignment) and execute independently through to a deliverable PR, with human review as a quality gate rather than a guidance requirement.
Strengths
- $1.5B valuation and $150M Series C (April 2026); $200M ARR run rate
- Enterprise customers include NVIDIA, Adobe, EY, Palo Alto Networks, MongoDB
- #1 on Terminal-Bench benchmark (58.75% score)
- LLM-agnostic: Claude, GPT-5, Gemini, DeepSeek per task
- 200% quarter-over-quarter growth driven by autonomous SDLC coverage
Limitations
- Enterprise-focused pricing; no self-serve free tier for individual developers
- Closed-source with proprietary Droid execution environment
- Full value realised only with deep Jira/Slack/GitHub integration — setup overhead
- Black-box execution model may be a concern for regulated industries requiring audit trails