Restless vs Paperclip.
Both are open source and both talk about running a company with AI agents. They start from different questions, and the difference matters more than the feature list.
Paperclip is probably the closest thing to Restless out there, so it'd be weird not to write this one. Both are open source, both let you organise AI agents into something like a company, and both care about goals, budgets and keeping humans in charge.
I'll try to be fair. Paperclip is a serious, popular project and it does a lot well. The difference is less about features and more about which question each one starts from.
What Paperclip is
Paperclip describes itself as the app people use to manage AI agents for work. You define a goal, then hire a team of agents into an org chart: a CEO, a CTO, engineers, marketers, each with a role, title and reporting line.
Work flows through issues linked back to projects and company goals, so agents can see why they're doing something. A CEO agent can create and delegate issues itself. Budgets at the company, agent and project level track spend, alert at thresholds and can pause work. Heartbeats wake agents when they have assigned work, follow-ups or scheduled routines.
It's MIT licensed, runs as a Node.js server with a React UI and an embedded Postgres, and it brings your own agent runtimes: Claude Code, Codex, OpenClaw, Cursor, Gemini CLI and a long list of others. It also supports human members, a ranked decision queue, versioned documents on issues and viewing artifacts.
Where Paperclip is great
I'd point you straight at Paperclip if any of these sound like you:
- You want to design and manage an agent organisation explicitly: roles, reporting lines, tasks and budgets.
- You already have favourite agent runtimes and want one place to orchestrate them all.
- You think in issues and tickets, and like delegation happening through a task system.
- You want a big, active open-source community around the project.
Where Restless is different
Restless starts from a different question: what useful work can we finish together, and where does human judgement make it better? The agents are there to carry the chores. The people stay in the work itself.
- You work on the result, not the ticket. Your cofounder edits the brief live, a colleague comments on a paragraph of the proposal, and you jump into the company computer to try the build.
- Attention is the budget. Routine coordination stays with the accountable lead. You get the output plus a prepared decision, not a queue of status changes.
- Teams stay small. One capable worker under each lead by default, with more only when they make the result better, check it independently or save time. I measured the cost of extra supervision and it wasn't pretty.
- The company knows itself. Approved facts, voice, visual language and ways of working travel with the work, and Restless tracks which version each output used. Change a fact and it flags what relied on the old one.
- Quality is a setting. Pick Fast, Thorough, Exceptional or Frontier for the company or a single outcome, then review the exact result.
Where they overlap
Plenty. Both have human members, goal-directed delegation, decision queues, budgets, document revisions and support for Codex and Claude Code. Both are self-hosted. If you're comparing checklists, they'll look similar.
The choice is really about the working relationship you want. Paperclip puts you in the seat of someone managing an agent organisation. Restless puts you in the room with the work, with agents doing the chores around you.
Honest trade-offs
Paperclip has a much bigger community and supports more agent runtimes out of the box. Restless is a development preview and its Claude and API-provider routes are newer than its Codex route. Restless also asks more of your machine, because each company gets a full persistent Linux computer rather than running inside your existing environment.
If you try both, I'd genuinely like to hear which one fit and why.