Deploy and Host Paperclip

Paperclip is an open-source AI agent orchestration platform that lets you run entire AI-powered businesses from a single dashboard. It organises Claude Code, OpenClaw, Codex, Cursor, and any heartbeat-capable agent into a structured company — complete with org charts, goal hierarchies, per-agent budgets, governance gates, and a full audit log.

Self-host Paperclip on Railway and get a production-ready deployment with a managed Postgres database pre-wired, persistent storage mounted at /paperclip, authenticated access enabled, and your public URL injected automatically — no manual configuration required.

The Paperclip service uses a custom Railway wrapper that adds a /setup onboarding page and a start.mjs entrypoint that handles database migration before the server starts.

Getting Started with Paperclip on Railway

Once your Railway deploy is live, navigate to /setup — this is the custom onboarding page included in the template. Click Register as Admin to create your admin account; this is the only account with full access to governance controls and billing. After registration, click Launch Paperclip to open the main dashboard.

Inside the dashboard, create your first company by clicking New Company. When prompted for a data directory, enter any path starting with /paperclip (for example, /paperclip/data) — this ensures your company files land on the mounted volume and survive redeploys. From there, write a mission statement for your company, add your first agents (CEO, CTO, engineers — any role you like), assign monthly token budgets, and click Go. Agents will begin operating on scheduled heartbeats from that point forward.

About Hosting Paperclip

Paperclip is a Node.js server with a React UI that models a business as an org chart of AI agents. It is MIT-licensed, self-hosted only (no cloud-hosted version currently exists), and requires no Paperclip account.

Key features:

• Org chart orchestration — assign CEO, CTO, engineer, and any other role to AI agents with defined reporting lines
• Goal-aware execution — every task carries full goal ancestry so agents know the "why", not just the title
• Per-agent budgets — monthly token caps enforced automatically; agents pause when they hit the limit
• Heartbeat scheduling — agents wake on a schedule, check their task queue, and act; event-based triggers also supported
• Multi-company isolation — one deployment can host dozens of companies with fully isolated data
• Full audit trail — every tool call, decision, and delegation logged in an append-only audit log
• Portable company templates — export/import orgs, agents, and skills with automatic secret scrubbing

Why Deploy Paperclip on Railway

Deploy Paperclip on Railway and get a production-grade AI orchestration platform without any infrastructure overhead:

• One-click deploy with Postgres pre-wired and migrations run automatically on startup
• Persistent volume mounted at /paperclip keeps company data safe across redeploys
• Private networking between the app and database — no public DB exposure
• Managed TLS and custom domain support out of the box
• Authenticated deployment mode enabled by default — no accidental public access
• Redeploy from Git with zero downtime whenever Paperclip ships a new release

Common Use Cases

• Solo founder running an AI content company — hire a writer, SEO analyst, and editor as agents; set the goal to "grow the blog to 100k monthly readers"; let Paperclip coordinate the work while you review strategy
• Developer building multiple AI ventures in parallel — spin up isolated companies for each project in one Paperclip instance; separate budgets, org charts, and audit logs per venture
• Agency managing client AI workflows — create a company per client; assign agents to client-specific goals; export/import org configs between client accounts with secret scrubbing
• AI researcher prototyping autonomous organisations — test multi-agent coordination patterns with real budget enforcement and goal tracing, not just prompt chaining

Dependencies for Paperclip

• Paperclip — Node.js 24, custom Railway wrapper (github.com/praveen-ks-2001/paperclip-railway-template), listens on port 3100
• Postgres — Railway-managed Postgres; connection injected via DATABASE_URL; stores all agent state, company data, sessions, and audit logs

Environment Variables Reference

Deployment Dependencies

• Runtime: Node.js 24
• Database: PostgreSQL (any version supported by Drizzle ORM)
• GitHub repo: github.com/paperclipai/paperclip
• Railway wrapper: github.com/praveen-ks-2001/paperclip-railway-template
• Official docs: paperclip.ing

Minimum Hardware Requirements for Paperclip

Paperclip vs Other Agent Orchestration Tools

Paperclip targets a different problem than CrewAI or LangGraph — it orchestrates an entire business, not just a pipeline. If you need to coordinate 20+ agents toward a shared mission with governance, budgets, and an org chart, Paperclip is the only self-hosted open-source tool that models this out of the box.

Self-Hosting Paperclip

Option 1 — Using the Railway wrapper (recommended for Railway):

git clone https://github.com/praveen-ks-2001/paperclip-railway-template.git
cd paperclip-railway-template
# Configure env vars, then deploy to Railway
railway up

Option 2 — Vanilla Docker with external Postgres:

docker run -d \
  -p 3100:3100 \
  -e DATABASE_URL="postgresql://user:pass@host:5432/paperclip" \
  -e BETTER_AUTH_SECRET="your-secret-here" \
  -e PAPERCLIP_PUBLIC_URL="https://your-domain.com" \
  -e PAPERCLIP_ALLOWED_HOSTNAMES="your-domain.com" \
  -e PAPERCLIP_HOME="/paperclip" \
  -v paperclip_data:/paperclip \
  --name paperclip \
  ghcr.io/paperclipai/paperclip:latest

For local development, the upstream npx paperclipai onboard --yes command spins up an embedded Postgres automatically — no external database required.

Is Paperclip Free?

Paperclip is fully open source under the MIT licence. There is no paid tier and no cloud-hosted version — self-hosted is the only deployment mode. On Railway, your cost is infrastructure only: the Paperclip service and the Postgres database. A typical low-traffic deployment runs well within Railway's Starter plan. Agent token spend is billed directly by your LLM provider (Anthropic, OpenAI, etc.) and is tracked per-agent inside Paperclip's dashboard.

FAQ

What is Paperclip? Paperclip is an open-source, self-hosted platform for running AI-powered businesses. It organises AI agents — Claude Code, OpenClaw, Codex, Cursor, HTTP bots — into a company with an org chart, goal hierarchy, per-agent budgets, and a full audit trail. You set the mission and govern from the top; the agents do the work.

Why must the company data directory start with /paperclip? The Railway template mounts a persistent volume at /paperclip. Any path inside that directory (e.g. /paperclip/data) will survive service redeploys and restarts. If you use a path outside /paperclip, company files will be written to the ephemeral container filesystem and lost on the next deploy.

Can I run multiple AI companies on paperclip? Yes. Paperclip supports unlimited companies in a single deployment with complete data isolation between them. Each company has its own org chart, agents, goals, budgets, and audit log. This makes it practical to run separate ventures, test different strategies in parallel, or manage client accounts from one Railway project.

Does Paperclip work with models other than Claude? Yes. Paperclip is agent-agnostic — any agent that can receive a heartbeat signal can be hired. Supported first-class adapters include Claude Code, OpenClaw, Codex, Cursor, OpenCode, and Pi. Custom HTTP-based agents and shell scripts also work. Set ANTHROPIC_API_KEY for Claude models or OPENAI_API_KEY for GPT models in the environment variables.

Is there a hosted/cloud version of Paperclip? No. As of now, Paperclip is self-hosted only. This Railway template is the fastest path to a production-ready cloud deployment.