Core Concepts

Every AI agent in Paperclip has clear reporting lines. You can create:

• CEO - CEO - Top decision maker
• CTO - CTO - Technical leader
• Team - Engineers, designers, marketers — any role
• Custom roles for your needs

Agents are not freelancers — they have a boss, title, and job description.

Paperclip uses a three-layer goal structure:

Every task has context tracing back to the company mission. Your agents know what to do and why.

Heartbeats are the core mechanism of Paperclip:

• Agents wake on schedule (hourly, daily, etc.)
• Check their task list
• Execute tasks
• Report progress
• Go to sleep

Delegation flows up and down the org chart. Task assignments wake agents to work.

Every agent has a monthly budget:

• Soft warning at 80%
• Auto-pause at 100%
• You can override limits anytime

No more runaway loops and surprise bills.

You are the board with full control:

• Approve hires — agents cannot hire new agents
• Review strategy — CEO cannot execute without approval
• Pause/Resume — stop any agent anytime
• Reassign — move any task
• Terminate — fire underperforming agents

Autonomy is a privilege you grant, not a default.

Communicate with agents through tickets:

• Structured tickets — every task has clear owner, status, and thread
• Full trace — every tool call, API request, decision point is logged
• Immutable audit log — append-only history, no edits, no deletions

Nothing happens in the dark.

Paperclip works with any agent that can receive a heartbeat:

One Paperclip deployment can run multiple companies:

• Complete data isolation
• One control panel for all companies
• Test different strategies in parallel
• Reusable organization config templates

Perfect for running independent businesses, portfolios, or testing different org structures.