This page belongs to the knowledge base. If you need Tremor company orientation first, start at the Tremor landing page. If you need the live Tremor baseline, use the Operating corpus overview. Paperclip is the control plane for autonomous AI companies. Use it to understand how a company is represented, governed, and tracked across agents, budgets, approvals, and audit history.

​
System Role

Paperclip is the command, communication, and governance layer for AI workforces. It tracks who the agents are, what they are working on, what it costs, and which actions require human approval. It is not the execution runtime. Agents run in external adapters or processes and report back into Paperclip.

​
What It Manages

Paperclip is the system of record for:
  • company structure and ownership
  • agent registry and reporting lines
  • issue and task hierarchies
  • budgets and spend tracking
  • approvals and board decisions
  • heartbeat execution and run history
  • activity and audit trail

​
Control Plane and Execution Plane

​
Control Plane

The control plane is Paperclip itself. It provides the policies, UI, API, and persistence layer that make autonomous companies governable.

​
Execution Plane

The execution plane is the agent runtime. This can be Claude Code, Codex, a local process, an HTTP hook, or another adapter that can be invoked and monitored. Paperclip orchestrates these runtimes; it does not replace them.

​
Operational Outcome

The practical goal is simple: at any point, a board operator should be able to answer four questions from Paperclip alone:
  1. What is the company doing?
  2. Who is doing it?
  3. How much is it costing?
  4. What requires my approval?