Paperclip: The Open-Source Platform Running Companies with AI Agents

What if you could run an entire company without hiring a single employee? Not a solo business where you do everything yourself—but a real organization with a CEO, engineers, marketers, and designers. All powered by AI agents working together toward shared goals. That's exactly what Paperclip delivers.

What is Paperclip?

Paperclip is an open-source orchestration platform that turns scattered AI agents into a functioning company. Built as a Node.js server with a React dashboard, it provides the organizational infrastructure that individual AI tools lack: org charts, budgets, governance, and goal alignment.

The key insight: The bottleneck wasn't agent intelligence—it was organizational structure. Paperclip solves this by modeling AI companies as real organizations with hierarchical reporting structures.

Why It's Going Viral

In March 2026, Paperclip hit 14,000+ GitHub stars in just one week. Here's why developers and entrepreneurs are paying attention:

• Open source & self-hosted: MIT licensed, no vendor lock-in
• Bring Your Own Agent: Works with Claude Code, OpenClaw, Codex, Cursor, or any agent that can receive a heartbeat
• Budget enforcement: Every agent gets a monthly limit—when they hit it, they stop automatically
• Multi-tenant: Run one AI company or fifty from a single deployment

How It Works

The Architecture

Paperclip uses a four-layer architecture:

1. Core Server: Node.js API running at localhost:3100
2. Database: Embedded PostgreSQL for local dev, external for production
3. Plugin Runtime: Isolated child processes communicating via JSON-RPC
4. React Dashboard: Monitor agents, costs, and governance in real-time

Goal Alignment

Context cascades from the top down:

Company Mission → Projects → Agent Goals → Individual Tasks

Every task traces back to the company's mission. Agents know what to do and why through inherited goal context—not just better prompts.

Heartbeat Scheduling

Agents don't run continuously. They wake on schedules, check assigned work, and take action. This keeps costs predictable and prevents runaway execution.

Key Features

Org Chart Model

Design your AI company like a real organization:

• CEO: Sets strategy, decomposes goals
• CTO: Technical direction, architecture decisions
• Engineers: Implementation, bug fixes
• Marketers: Content, campaigns
• Designers: UI/UX, visuals

Every agent (except CEO) reports to a manager, creating clear accountability.

Budget Control

Set monthly budgets per agent. When an agent hits its limit, it stops—no surprise bills, no runaway costs. Track spending per agent, per task, per project, per goal.

Governance & Approvals

• Approval gates for sensitive operations
• Immutable audit logs
• Full tool-call tracing
• Board approval workflows

Pre-Built Companies

The paperclipai/companies repo includes 16 pre-built company templates with 440+ specialized agents and 500+ skills:

• Security auditors
• Game studios
• Research labs
• Dev shops
• Marketing teams
• Content agencies

Getting Started

The fastest way to spin up Paperclip:

npx paperclipai onboard --yes

This launches the app in your browser and walks you through:

1. Setting up your first AI CEO
2. Approving the hire of specialists
3. Configuring budgets and goals
4. Watching your "company" start working

For production deployment, point it at your own PostgreSQL instance and deploy however you like—Vercel, Zeabur, or your own infrastructure.

Real-World Example: Content Team

Here's how an 11-agent content team might work:

Each agent has a specific role, budget, and reporting line. The CEO coordinates the team toward the company's content goals.

The Ecosystem

Paperclip is rapidly building an ecosystem:

• Hermes Adapter: Run Nous Research's Hermes Agent (30+ tools, 80+ skills, MCP support) as a managed employee
• Desktop Companion: Monitor AI coding agents without staring at your terminal
• CLI Tools: paperclipper for bootstrapping full companies
• Tutorial Platform: 7-day structured course for building zero-human companies

Challenges to Consider

Paperclip isn't a magic bullet. Some challenges:

• Setup complexity: Initial configuration requires technical knowledge
• Goal decomposition: Breaking down high-level goals into agent tasks takes iteration
• Cost monitoring: Even with budgets, tracking ROI per agent requires attention
• Agent coordination: Cross-team requests need careful delegation logic

The platform is also young—just a few weeks old at the time of writing. Expect rapid changes and some rough edges.

The Bigger Picture

Paperclip represents a shift in how we think about AI agents:

Old paradigm: One powerful agent doing everything New paradigm: Specialized agents organized like a company

This mirrors how human organizations work. We don't expect one person to be CEO, engineer, marketer, and designer. We build teams with specialized roles. Paperclip applies the same logic to AI.

Conclusion

Paperclip is the infrastructure layer that AI agents have been missing. It doesn't make agents smarter—it makes them organized. With org charts, budgets, governance, and goal alignment, you can run a coordinated AI workforce instead of managing isolated tools.

The 14,000+ GitHub stars in week one suggest this resonates. Whether you're a solo founder looking to scale, a developer building agent workflows, or just curious about the future of work—Paperclip is worth exploring.

Try it: npx paperclipai onboard --yes