Software factory platforms.

• Where does work come from?A ticket, an issue, a chat message, a Slack thread or a CI failure. The intake decides who can hand the line work.
• How is each unit isolated?Its own branch, worktree, container or cloud sandbox, so parallel runs do not step on each other.
• What checks run before you look?Tests, builds and linters the agent runs itself, with the results attached to the change.
• What does review look like?A pull request, a diff in a dashboard or a card in a column. Review is where the hours go once code is cheap.
• Whose machine and whose model?Local, cloud or both, and whether you bring your own AI subscription or pay per use.

New to the term? What a software factory is, and how one works

1. Cursor Cloud Agents

   Cursor's agents that run in isolated cloud VMs with a full development environment, building and testing a feature for you to review. Formerly called Background Agents, they start from the desktop app, the web, Slack, GitHub comments, Linear or an API. Suits: Teams already working in Cursor who want to hand off tasks and stay on the decisions. Cursor Cloud Agents vs Fredrin

   Cloud agents from an editor

2. Devin

   Cognition's autonomous agent, which plans, writes and tests code in the cloud or on your machine, using its own browser, and works until a pull request is ready to merge. Suits: Teams that want to delegate whole tasks to an agent that works on its own. Devin vs Fredrin

   Autonomous software engineer

3. Factory

   An agent-native software development platform whose agents, called Droids, plan, build, review, test and ship. They run from the command line, desktop, web, Slack, Microsoft Teams, Jira, Linear and CI. Suits: Engineering teams that want to delegate backlog work, pull request review, tests and incident investigation without replacing their existing tools.

   Agent-native development platform

4. Fredrin

   A desktop kanban board where every ticket gets its own branch, git worktree and agent session, running Claude Code or Codex on the subscription you already pay for. Finished tickets land in Review with their diff and checks. Suits: A team, or one person, keeping many tickets moving at once and reviewing the output together.

   Multiplayer board for AI Workers

5. GitHub Copilot cloud agent

   GitHub's agent that picks up an issue and returns a pull request, and can plan and make changes, review pull requests and automate repository tasks. GitHub's docs say it is being folded into a new experience, Copilot in the cloud. Suits: Teams on GitHub with a paid Copilot plan who want agents working from their issues.

   Issue-to-pull-request agent

6. Nx

   A monorepo build platform that runs only what a change affects. Nx Cloud adds Self-Healing CI, which detects, analyzes and proposes fixes for CI failures. Suits: Developers and platform teams working in a monorepo who want AI help with CI.

   Monorepo platform with self-healing CI

7. OpenAI Codex

   OpenAI's coding agent, available as the Codex CLI, an IDE extension and Codex Cloud, and inside the ChatGPT desktop, web and mobile apps. Suits: Developers who want one agent across their terminal, editor and the cloud.

   Coding agent

8. Opsera

   A platform for governed software delivery, describing itself as a context-aware software factory that turns intent into working software through a repeatable, governed process. Suits: Large engineering, data and IT organizations that need enterprise control over AI-assisted delivery.

   Governed software delivery

9. Warp

   An agentic development environment whose Oz cloud agent platform and Warp Factories run fleets of coding agents across the software lifecycle, with the setup defined in code. Suits: Developers and engineering teams who want to define and run fleets of cloud agents.

   Agentic development platform