Paperclip

Learning Objectives

• Understand what Paperclip is and how it differs from other AI agent orchestration frameworks
• Identify Paperclip's core concepts: org charts, heartbeats, budgets, governance, and multi-company support
• Evaluate when Paperclip is the right choice for autonomous agent workflows

What Is Paperclip?

Paperclip is an open-source orchestration platform that treats AI agents like employees in a company. Instead of managing individual agent prompts and chaining workflows manually, Paperclip lets you build an organizational chart where agents have defined roles, reporting lines, budgets, and goals — enabling fully autonomous business operations without human management overhead.

Founded in 2025, Paperclip has quickly gained traction in the AI community with over 37,000 GitHub stars. The platform is built in TypeScript with a React interface and supports any agent type, including Claude, OpenClaw, and Cursor.

Core Concepts

Org Chart for Agents

Paperclip's central metaphor is the company org chart. Instead of defining agent pipelines or chains, you:

• Hire agents into specific roles (CEO, developer, marketer, QA engineer)
• Set reporting lines so agents know who to escalate to and who to delegate to
• Define responsibilities for each role, just like a real job description
• Assign goals that trace back to the company's overall mission

This approach makes complex multi-agent systems intuitive — the organizational structure itself becomes the orchestration logic.

Heartbeats — Scheduled Agent Activation

Heartbeats are Paperclip's scheduling system. Instead of waiting for triggers or running agents continuously, heartbeats activate agents on a defined schedule:

• A marketing agent might have a daily heartbeat to review analytics and draft social posts
• A QA agent could run hourly to scan for new pull requests and review code
• A CEO agent might check in weekly to evaluate company-wide progress against goals

Heartbeats keep agent costs predictable while ensuring work gets done on schedule.

Cost Controls and Budgets

Every agent in Paperclip has a monthly budget that is automatically enforced:

• Set spending limits per agent to prevent runaway API costs
• Track token usage and API calls in real time
• Agents automatically pause when they hit their budget ceiling
• Budget reports show which agents are delivering value and which are overspending

Governance Layer

For high-stakes decisions, Paperclip provides a board-level governance system:

• Define which decisions require board approval before execution
• Set approval thresholds (majority, unanimous, specific board members)
• All governance decisions are logged with full rationale and audit trails
• Prevents autonomous agents from making costly or irreversible mistakes without oversight

Ticket System and Audit Trails

Every action in Paperclip is tracked through a ticket system:

• Agents create tickets for tasks, decisions, and deliverables
• Full decision logging captures why an agent took a specific action
• Audit trails make it easy to debug agent behavior and trace outcomes back to goals
• Ticket history provides accountability across the entire agent workforce

Key Capabilities

Getting Started

Paperclip can be set up in minutes with its interactive CLI — run npx paperclipai onboard --yes to get started.

This command walks you through:

1. Setting up a database — embedded Postgres locally or a custom connection
2. Creating your first company — name, mission, and initial goals
3. Hiring your first agents — assign roles and connect to LLM providers
4. Configuring heartbeats — set activation schedules for each agent
5. Setting budgets — define monthly spending limits per agent

The platform also offers pre-built company templates through its companies repository, so you can import proven agent team configurations for common use cases like software development, marketing, or customer support.

Use Cases

• Autonomous software development — hire a team of developer, QA, and DevOps agents that write, test, and deploy code
• Content and marketing operations — agents that research topics, draft content, schedule posts, and analyze engagement
• Customer support automation — tiered agent support with escalation to senior agents for complex issues
• Research and analysis — teams of agents that gather data, synthesize findings, and produce reports
• Multi-business management — run multiple independent agent-powered businesses from one Paperclip instance

Limitations

• Self-hosted only — no managed cloud offering yet; requires infrastructure to run
• Early stage — rapid development means APIs and features may change between releases
• LLM costs — while Paperclip is free, the AI agents it orchestrates still incur API costs from their LLM providers
• Learning curve — the org chart metaphor is intuitive but requires rethinking traditional automation approaches
• Community-driven — documentation and ecosystem are still growing

How Paperclip Compares

Related Tools

• CrewAI — role-based multi-agent framework
• AutoGen — Microsoft's multi-agent conversation framework
• n8n — open-source workflow automation with AI agent nodes
• LangChain — LLM application development framework
• OpenAI Agents SDK — OpenAI's agent building toolkit