March 31, 2026
Agentbot: The Managed Platform for Self-Hosted AI Agents
Your AI agent. Your hardware. Your rules.
Today we're launching Agentbot — a managed platform for deploying autonomous AI agents on your own hardware. One command. Your API key. Your channels. Your data.
The Problem
Cloud AI tools log your data, mark up your API costs, and lock you in. You're renting intelligence you don't control.
ChatGPT, Claude, and every "AI assistant" SaaS share the same model: your data goes to their servers, they charge you a premium on top of the LLM provider's cost, and switching providers means losing everything.
That's not a product. That's a trap.
The Solution
Agentbot puts the agent on your hardware.
curl agentbot.raveculture.xyz/install | bash
That's it. One command. Your agent is running on your machine — connected to Telegram, Discord, or WhatsApp. Your API key talks directly to your LLM provider. We don't touch the costs.
What You Get
- Self-hosted by design. Your data never leaves your machine. No cloud dependency. No vendor lock-in. Run it on a Raspberry Pi, a Mac Mini, or a VPS — wherever you have Docker.
- Multi-channel out of the box. Connect your agent to Telegram, Discord, and WhatsApp simultaneously. One agent, multiple channels, unified memory.
- BYOK — Bring Your Own Key. Use OpenAI, Anthropic, Google, or any OpenRouter-compatible provider. You pay them directly. Zero markup from us.
- Autonomous agents, not chatbots. Memory that persists across conversations. Skills that compound. Scheduled tasks that run while you sleep. This isn't a decision tree — it's a thinking machine.
- Onchain-native payments. x402 settlement on Base. Agent-to-agent marketplace. $AGENTBOT token for governance. The agent economy isn't coming — it's here.
- Open source core. The architecture is MIT-licensed on GitHub. Inspect the code. Fork it. Build with us.
How It Works
- Install — Run one command on your machine
- Configure — Set your API key and connect your channels
- Deploy — Your agent is live in 60 seconds
- Customize — Add skills, set memory, schedule tasks
No YAML. No config files. No DevOps degree required.
Pricing
We don't do free tiers. Everyone pays. This is a feature — it filters out noise and signals quality.
- Solo: £29/mo — 1 agent, personal use
- Collective: £69/mo — 3 agents, small team
- Label: £149/mo — 10 agents, white-label for clients
- Network: £499/mo — Unlimited agents, enterprise
The Stack
- Frontend: Next.js 16 (Vercel)
- Backend: Node.js + Docker (Render)
- Database: PostgreSQL + Redis
- Payments: Stripe + x402 on Base
- Auth: NextAuth (GitHub, Google, Wallet/SIWE)
- Agent Runtime: OpenClaw (Docker)
Security
AI agents handle sensitive data. We took security seriously from day one:
- Edge auth middleware (NextAuth session gate)
- Rate limiting on all auth endpoints
- AES-256-GCM token encryption at rest
- HMAC-signed OAuth state parameters
- 0 npm vulnerabilities
- Webhook signature verification (Stripe, Resend, Mux)
- SSRF protection on all external API calls
The Vision
The agent economy is here. Agents that run on your hardware. Pay with x402 on Base. Memory that persists. Skills that compound.
We're not building a chatbot wrapper. We're building the infrastructure layer for autonomous AI agents.
Get Started
→ Docs
Your AI agent. Your hardware. Your rules.