Paperclip
Paperclip is an open-source, self-hosted app for managing a team of AI agents for work, with org charts, goals, tasks, budgets, and governance in one deployment.
About Paperclip
Paperclip
Paperclip lets users organize AI agents into a company-style structure. Users define a goal, hire agents such as a CEO, CTO, engineers, or marketers from any provider, and approve strategy and budgets before running them. The page describes features including bring-your-own-agent support, org charts, goal alignment, scheduled heartbeats, per-agent monthly budgets with automatic stops, ticket-based tracing and audit logs, and governance controls where the user acts as the board. It is MIT-licensed, can run locally with an embedded Postgres database or against your own Postgres, and supports multiple companies within one deployment. A waitlist and local install are offered.
Key features
- Bring Your Own Agent: works with any agent runtime that can receive a heartbeat
- Org chart with hierarchies, roles, and reporting lines
- Goal alignment linking tasks to company mission
- Heartbeats that schedule agents to check work and act
- Monthly budgets per agent that pause agents at limit
- Cost tracking per agent, task, project, and goal
- Ticket system with full tool-call tracing and immutable audit log
- Governance with board approval for hires, pause, override, reassign, and terminate controls
- Open source under MIT license, extensible through extensions
- Support for running multiple companies in one deployment with data isolation
- Local or remote deployment with interactive setup via npx paperclipai onboard
Best suited for
- Teams running multiple AI agents across business functions such as dev, content, social, marketing, QA, research, and outreach
- Builders who want self-hosted, open-source orchestration of agent workforces
- Users who want cost limits and governance over autonomous agents
- Operators running separate ventures or testing strategies in parallel
Pricing
Not stated on the website
Discussion
Used Paperclip? Share what worked and what didn’t.
No comments yet. Start the conversation.