Agent orchestration
Paperclip
The company, not just the employee
Our verdict
The only serious open-source platform for orchestrating teams of agents. If you're past one agent doing one thing, Paperclip is the layer you need.
Best for
Teams running multiple AI agents who need org structure, budget controls, and approval workflows across their agent workforce.
Not for
Anyone just getting started. Paperclip is infrastructure, not an entry point.
Overview
Paperclip's own framing says it best: 'If OpenClaw is an employee, Paperclip is the company.' It's an orchestration layer — org charts, reporting lines, budget limits, approval gates, audit logs — for a workforce of AI agents. It works with any agent runtime: OpenClaw, Claude Code, Cursor, custom HTTP agents. The heartbeat system wakes agents on schedule, assigns tasks, tracks what they spent, and surfaces anything that needs human review. Hard per-agent budget limits prevent runaway API spend. The full audit trail with immutable decision history means you can reconstruct exactly what an agent did and why. Latest release was April 16, 2026.
Repository activity
Updated 2 days ago
Stars
69,721
+10,155 in 45d
Forks
12,938
Contributors
110
Last release
v2026.529.012 days ago
Last commit
3 days ago
What works
- +The only open-source multi-agent orchestration platform
- +Works with any agent runtime — fully vendor-agnostic
- +Hard budget limits per agent prevent runaway API costs
- +Immutable audit trail for every agent decision
- +Active development — latest release April 2026
What doesn't
- −Not a starting point — assumes you have agents to orchestrate
- −Self-hosting requires PostgreSQL and Node.js infrastructure
- −Smaller community than OpenClaw or Hermes
What operators use it for
01
Building and Launching SaaS Products
Paperclip can run an entire product team autonomously. One user built a company called Proof Shot — a tool for automating customer testimonials — entirely inside Paperclip. A CEO agent hired a founding engineer and a QA agent. Those agents built the backend, the AI processing pipeline, and an embeddable frontend UI. No human wrote the first line of code.
02
Full-Stack Newsletter Automation
Stand up a four-agent team — researcher, writer, editor, reviewer — and your newsletter runs itself. The team pulls performance metrics from previous issues, brainstorms angles, drafts the content, and pushes updates to ConvertKit. Each cycle, it's working from data on what actually performed. Growth becomes a system, not a task.
03
Complex Lead Generation and Data Aggregation
Paperclip handles research workflows that require combining multiple data sources into a single judgment. One roofing company uses it to identify high-probability leads by cross-referencing satellite imagery with recent hail storm data in high-income neighbourhoods. The agent team surfaces the list. The sales team works it.
04
Automated Security Reviews and Code Audits
Schedule a nightly security agent to sweep your codebase for exposed secrets, API keys, and known vulnerability patterns. Existing cybersecurity firms are already running Paperclip to deliver automated audit reports to their clients — the same checks, every night, without a consultant on the clock.
05
Autonomous Video Game Development
Paperclip ships with pre-built company templates. Import a game studio template and you have a creative director, producer, and technical director ready to take a brief. One team used this setup to build a bullet-hell game inspired by Vampire Survivors in Godot — the agents wrote the design doc, built the game, and iterated on it without a human in the loop.
06
Open-Source Community Management
Run a daily agent that monitors your GitHub repo, reads every pull request merged in the last 24 hours, drafts a Discord message calling out the contributors by name, and posts it automatically. Your community sees recognition without you managing the process. Consistency without overhead.
07
Scaling Content Strategy and Marketing
Assign agents to your marketing function and set a brief. They can develop a LinkedIn content calendar, produce social media carousels, or plan and script 60-second promotional videos using tools like Remotion. The output isn't a plan — it's the actual assets, ready for review and publish.
08
UI/UX Design Review and Bug Fixing
Upload a screenshot of a broken UI element — a misaligned button, a pill that's wrapping wrong — and assign it to a coder agent with a single instruction. The agent fixes the code and passes it to a QA agent with browser access, which visually confirms the fix before closing the ticket. Design PM work, without the back-and-forth.
09
Dedicated Scientific Research
Paperclip supports importing large pre-built agent architectures — teams of up to 48 agents configured specifically for scientific research. These teams come with structured knowledge bases and role definitions modelled on real doctors and scientists. You describe the research question. They work it.
10
Foundation and Family Organisation
Paperclip's delegation model applies anywhere you have tasks, roles, and output to track — not just software. A dentist currently runs a Paperclip agent company to manage their family foundation and handle day-to-day household administration. The same structure that runs a SaaS product team runs a family calendar and grant process.
Pricing
Free and open-source. Self-hosted on Node.js + PostgreSQL.
Common questions about Paperclip
What is Paperclip?
Paperclip is an open-source orchestration layer for AI agents — org charts, reporting lines, budget limits, approval gates, and audit logs for a workforce of agents. It works with any agent runtime (OpenClaw, Claude Code, Cursor, custom HTTP agents). Often described as 'if OpenClaw is an employee, Paperclip is the company.'
Is Paperclip free?
Paperclip itself is free and open-source. You pay for the underlying model API costs the agents consume. Total monthly costs for solo operators typically land between $50–$300 depending on how many agents run and what models they use.
How does the 'zero-human company' pattern work in Paperclip?
You define an org chart of agent personas (CEO agent, CTO agent, QA agent, marketing agent), give each a budget and a role, and the orchestrator schedules their work and routes tasks between them. Humans act as the board of directors — setting high-level goals and reviewing output rather than driving each step. The pattern works at small scale (5–15 agents) and starts to break around 30+ where orchestration overhead eats the per-agent gain.
Paperclip vs CrewAI vs LangGraph?
Different layers of the stack. CrewAI and LangGraph are frameworks you write code in to build agent workflows. Paperclip is a platform you run that operates agents already built. If you're a developer wanting fine-grained control, CrewAI or LangGraph. If you're an operator wanting to manage multiple agents with budgets, approvals, and audit logs, Paperclip.
Continue exploring
Tool
AI Agent Cost Calculator
See what running Paperclip's typical workflow costs at your volume — pre-filled for this category.
Tool
The AI Agent Picker
Not sure Paperclip is right? Five questions, one recommendation.
Article
What Claude Skills actually are
Most builders think Skills are saved prompts. The architecture is different.
More in Open-source harnesses
Open dataset. This review is part of a structured dataset of every platform on the shortlist, published as platforms.json on GitHub under CC-BY-4.0.