Factory

Open original

Factory is a marketplace listing for an enterprise platform that uses Claude-powered "Droids" to plan, write, review, test, and document software, and respond to incidents, with human approvals and audit trails. It runs across the terminal, IDE, web, desktop, and CI/CD, and can be bought through the Claude Marketplace.

  • Droids handle the full software lifecycle, from planning and coding through review, testing, documentation, and incident response, using context pulled from repositories and connected tools.
  • Governance is built in: approvals, policies, and audit trails are meant to keep humans in control of autonomous work.
  • It integrates with Jira, Notion, Slack, and GitHub and can deploy locally, in the cloud, or in CI/CD. Existing Anthropic commitments can be applied to purchase.

Factory is an enterprise product that uses Claude-powered agents, which the page calls "Droids," to handle software engineering work autonomously. The pitch is that these agents read a company's codebases and then carry out tasks across planning, coding, code review, testing, documentation, and incident response. They can run in a terminal, an IDE, a web interface, a desktop app, or inside CI/CD pipelines, and they pull context from repositories and connected tools. The stated goal is to let enterprises keep building and operating production software with less hands-on human work at each step. On the practical side, the listed capabilities are fairly concrete. Teams can hand off whole features or bug fixes and get back tested implementations, rather than just code suggestions. The product integrates with Jira, Notion, Slack, and GitHub, and can be deployed locally, in the cloud, or inside a build pipeline. The page also emphasizes governance: approvals, policies, and audit trails are meant to keep the agents within boundaries a company sets. That last part matters for anyone thinking about giving autonomous tools write access to production code. For purchasing, Factory is sold through the Claude Marketplace, and existing Anthropic commitments can be applied toward it. Beyond that, the page gives no pricing, no customer names, no performance numbers, and no detail on how the testing or approval workflows actually function. It also carries a standard disclaimer that Anthropic does not guarantee the provider's content is accurate. So the claims here are the vendor's own, and anyone evaluating the product would need to verify them through a trial or direct conversation with the company.

Ask about this article