Belong.net Logo
Belong AI Deploy

Quick Start

Choose the guided hosted flow or self-host from the public aideploy repository.

Quick Start

Pick one setup path. Both run the agent in your cloud account; the difference is who operates the deployment layer.

PathBest forStart here
Guided hosted setupFast launch, more providers and channels, managed operationsaideploy.co
Public self-host betaSource control, local lifecycle CLI, DigitalOcean + Telegram golden pathbelongnet/aideploy

Guided hosted setup

  1. Go to aideploy.co
  2. Sign in with DigitalOcean — click the button, authorize, done (~10 seconds)
  3. Sign in with ChatGPT — click the button, authorize, done (~10 seconds)
  4. Connect Telegram — paste your BotFather token (~30 seconds)
  5. Click "Launch My Agent" — wait for the progress bar to finish (~60–90 seconds)
  6. Done! — your agent is live. Send it a message on Telegram.

What you need

  • A DigitalOcean account (credit card required for billing; eligible new accounts can receive a $200 / 60-day referral credit)
  • A ChatGPT account (Plus or Pro recommended)
  • A Telegram account
  • A Tailscale account (free) for private dashboard access

You can swap any of these for another cloud provider, AI provider, or messaging channel.

Open-source self-host setup

The public beta supports:

  • DigitalOcean as the live cloud path
  • OpenClaw or Hermes Agent
  • OpenAI, Anthropic, or Kimi with an API key
  • Telegram with outbound polling
  • A Tailscale-private browser surface

Start with the repository's self-host guide. It gives you the current tagged-source command, required credentials, checksum-pinned OpenTofu setup, readiness checks, and deterministic teardown. The short lifecycle is:

node cli/dist/index.js up
node cli/dist/index.js doctor
node cli/dist/index.js down <deploy-id>

The public CLI validates every prerequisite before creating billable resources. It records state under ~/.aideploy/<deploy-id>/, reports when the real runtime and Telegram path are ready, and removes secrets and infrastructure during a successful teardown.

See Open Source for the public-versus-managed feature boundary, licensing, release verification, and contribution links.

After deployment

  • Open your dashboard — reach the management dashboard over your Tailscale private network (:3000), select your agent, then go to Connect AI & Messengers.
  • Finish setup from chat — once the bot is replying in Telegram, you can finish any remaining provider or channel setup right there. Ask it to connect ChatGPT or Claude, open the sign-in link it sends, then paste the code back to the bot. It deletes that sensitive message when possible.
  • Add more agents — go to Agents and click New Agent to deploy additional agents.
  • Install Tailscale — add the Tailscale app to your phone or laptop to reach the dashboard from anywhere.
  • Test it — confirm Telegram is connected in the setup checklist, then send a message to your bot.

Estimated costs

  • DigitalOcean server: ~$24–30/month (Starter plan, 2 vCPU / 4 GB RAM)
  • ChatGPT: included in your existing subscription when you sign in with OAuth
  • Tailscale: free for personal use

Security note

Never paste API keys or OAuth bearer tokens directly into chat.

When you finish ChatGPT or Claude sign-in from a messenger, the bot sends a browser sign-in link. Copy that link into Safari, Chrome, or your normal desktop browser — messenger in-app browsers cannot finish the login reliably. Paste back the ChatGPT sign-in code (or the Claude redirect URL / one-time code) the browser shows you.

For raw provider API keys, use the dashboard's secure setup forms instead of pasting keys into chat.

Need help?

  • Check the Troubleshooting guide
  • Review the deployment logs in the Technical details section of the wizard
  • For source installs, run node cli/dist/index.js doctor and use the repository's issue tracker
Copyright © 2026