Paperclip

Paperclip is a platform that lets organizations run multiple AI agents as a coordinated team. It provides an open-source control plane where each agent appears on a single org chart, owns tasks tied to goals, and operates within budgets. The system supports human review stages and auditability, and it is designed for teams that manage AI agents across different functions and runtimes.

Paperclip operates as a management layer rather than a model or runtime. It tracks decisions within tasks, ties work back to goals, and maintains a history of communications, approvals, and costs. The platform supports agents from multiple runtimes and can integrate with various execution environments.

It emphasizes tickets, review gates, cost controls, and audit logs to ensure accountability. It also highlights features like role-based access, delegation, and the ability to connect adapters for different agent runtimes. The documentation frames Paperclip as a platform for running a company of agents rather than a single tool.

Paperclip presents plans to install and run in various environments, including local, embedded database, and cloud deployment. It emphasizes self-hosting, MIT-licensed open source, and a cross-provider runtime that supports any model and any agent. The connectors page lists many integrations and adapters to various apps, indicating broad platform compatibility and an extensible architecture.