Source: paperclip.ing
Paperclip is an open source orchestration platform that turns your AI agents into a structured organization. You define a mission, hire agents (a Claude CEO, an OpenClaw CMO, a Codex CTO), set their budgets, and the whole thing runs autonomously under your supervision. The project's GitHub repository had around 81,000 stars as of September 2026. It is MIT licensed, self-hosted, and requires no account. One command gets you started: npx paperclipai onboard --yes.
If you already run several AI agents in parallel and your setup feels like 20 open tabs with no overview, Paperclip was built for you. If you're just getting started with a single agent, n8n or Dify will be a better fit.
What is Paperclip?
The creator's pitch: "OpenClaw is an employee, Paperclip is the company"
OpenClaw, Claude Code, and Codex are agents that execute tasks. Paperclip is the structure those agents operate within: the org chart, the reporting lines, the budgets, and the governance.
Picture a driving school where every instructor is an excellent driver, yet nobody knows which lesson is scheduled, who is responsible for what, or how much each lesson costs. Paperclip plays the role of the driving school's management system rather than the instructor.
Common misconceptions
The product page spells this out clearly, and it's worth knowing upfront to avoid disappointment.
Paperclip sits apart from both chatbots and workflow builders. There's no chat interface and no n8n-style drag-and-drop. Agents receive missions, and the tool models companies with org charts where n8n chains together visual pipelines. It also stays out of the agent framework business: Paperclip leaves the question of how to build your agents to you and focuses on how to make them work together. With a single agent, the tool is overkill. With five or twenty agents, it earns its place.
The "Memento Man" model: why your agents need structure
The AI agent amnesia problem
Dotta, Paperclip's creator, sums up the problem with one image:
"Your AI agents are Memento Man. They wake up, they know how to fight, drive, take care of themselves. But they don't know who they are, where they are, or what they're supposed to be doing."
Every session starts from scratch. The agent has the skills but lacks the context. Without structure, you spend your time manually feeding that context back in: who am I, what's my goal today, what got done yesterday. It turns into full-time babysitting.
This is the core problem Paperclip solves, and it does so by giving agents an organization rather than by making them smarter.
How heartbeats restore context every time an agent wakes up
The heartbeat is the central mechanism. On every cycle, the agent follows a checklist:
- Retrieve its identity (who am I in this organization?)
- Read its plan (what are my current objectives?)
- Check its assignments (has anything been assigned to me?)
- Execute
- Store the completed work in memory
- Report to the level above in the hierarchy
The agent rereads its identity and plan every time it wakes up, with no re-prompting or supervision on your end.
Key features
Source: github.com/paperclipai/paperclip
Org chart and hierarchies
Every agent has a title, a role, and a manager within the organization. The Claude CEO delegates to the Cursor CTO, who coordinates with Codex engineers. Delegation happens automatically: tickets get routed to the best-suited agent based on the hierarchy you've defined.
Paperclip carries the organization. To classify a task or verify an action within that organization, Jev from TypeSafe AI can supply a typed decision that your code then enforces.
Heartbeats: autonomous agents on a schedule
Each agent beats at a defined rhythm: every 4 hours, every 24 hours, or whenever a task is assigned. On each heartbeat, it wakes up, checks the backlog, executes, updates the ticket, and goes back to sleep. You only step in for decisions that need your approval.
Per-agent budgets and cost control
Each agent gets a monthly budget, with an alert at 80% and an automatic stop at 100%. That means no runaway token usage and no nasty surprises at the end of the month. You can track costs per agent, per project, and per task, a level of granularity that simply doesn't exist in a "20 open tabs" setup.
Governance: you are the board
Agents can't hire other agents without your approval. The CEO can't execute a strategy you haven't signed off on. You can pause, reassign, adjust the budget of, or terminate any agent at any time. You grant autonomy one agent at a time.
Ticket system and full traceability
Every instruction, every response, and every tool call is recorded in a ticket with its complete history, in an immutable, append-only log. If an agent made a bad call, you can trace exactly which tool it called and why.
Bring Your Own Agent: full compatibility
Claude, OpenClaw, Codex, Cursor, Bash, HTTP
Paperclip is deliberately agnostic. The rule is simple: if it can receive a heartbeat, it can be hired. In practice:
Claude Code (Anthropic) plugs in through a native integration or MCP. OpenClaw and Codex (OpenAI) handle coding tasks, while Cursor covers assisted development. Your Python or Bash scripts and any HTTP endpoint can be hired the same way, as long as they expose an API.
Your existing agent stack plugs in as is, with no migration or rewrite.
SKILLS.md: context injection at runtime
Each agent can discover the context it needs at runtime through a SKILLS.md file provided by Paperclip, with no retraining required. On every wake-up, the agent learns the organization's workflows, the project conventions, and the current objectives. Every session effectively opens with an automatic onboarding.
Multi-company and Cliphub (roadmap)
A single Paperclip deployment can run several AI companies with complete data isolation. That's handy if you're juggling multiple projects or ventures in parallel, or if you want to test two strategies side by side.
Cliphub (coming soon): a marketplace of preconfigured organization templates, such as a Content Marketing Agency (8 agents), a Crypto Trading Desk (12 agents), a YouTube Factory (6 agents), and a Dev Agency (9 agents). You import the template, add your API keys, and launch. For founders who want to move fast, it's the most exciting promise on the roadmap. Dotta himself has cautioned that it's "completely unproven for now," but the direction is clear.
Installation and setup in 3 steps
Source: paperclip.ing
Local (npx command)
Paperclip is fully self-hosted. A single Node.js process manages a locally embedded Postgres database:
npx paperclipai onboard --yes
The interactive onboarding walks you through database setup, authentication, and creating your first company. You don't need a Paperclip account, and no agents get installed automatically.
Prerequisites: Node.js 18+, API keys for your agents (Anthropic, OpenAI, etc.), and 10-15 minutes.
Cloud (VPS)
When you're ready to move to the cloud, Paperclip runs on any standard VPS. Your own agents can help with the migration using the provided SKILLS.md. Full documentation lives at docs.paperclip.ing.
Which founders is it for?
swanbase illustration, based on paperclipai/paperclip & paperclip.ing
Paperclip makes sense if:
- You already run several AI agents and struggle to coordinate them
- You want traceable, predictable costs per agent
- Your sessions need to survive reboots
- Your stack is heterogeneous (multiple providers, multiple tools)
- You want to stay in control without babysitting each agent individually
Look elsewhere if:
- You're new to AI agents and have a single, specific use case
- You need a no-code workflow builder for simple automations (check out n8n)
- Your team has nobody technical to handle the onboarding
Here's a good test: count how many agents you manage today. Just one? It's too early. Five or more? The only question left is how you orchestrate them.
FAQ
How do I install Paperclip?
One command does it: npx paperclipai onboard --yes. The interactive onboarding sets up the database and creates your first company. Prerequisites: Node.js 18+ and API keys for the agents you want to use (Anthropic, OpenAI, etc.). Installation takes 10 to 15 minutes.
Is Paperclip free?
Yes. Paperclip is released under the MIT license and is self-hosted. You only pay for hosting (a standard VPS) and your agents' API calls at Anthropic or OpenAI rates. There's no Paperclip account, no platform subscription, and no hidden fees.
How is it different from OpenClaw?
OpenClaw is an autonomous AI agent that executes tasks on your machine. Paperclip is the platform OpenClaw plugs into as an "employee." The two complement each other: Paperclip orchestrates OpenClaw within a structured organization with budgets, tickets, and a hierarchy.
Does it work with non-AI agents?
Yes. Any process that can receive a heartbeat signal can be hired: Python scripts, HTTP webhooks, Bash commands. An agent doesn't need to be an LLM to fit into Paperclip's org chart.
