I'm João, a solo developer from Portugal, and I just launched OpenClaw Cloud — a managed hosting platform that gives you your own personal AI assistant running 24/7 in the cloud, connected to all your chat apps.
The Problem
I've been running my own AI assistant for a while — a bot that connects to WhatsApp, Discord, Telegram, Slack, and more. It's incredibly useful: I delegate tasks, ask questions, get summaries, and it's always available.
But running it required maintaining a server, handling updates, managing Docker containers, and keeping everything alive. Most people who'd benefit from a personal AI assistant don't want to deal with that.
The Solution
OpenClaw Cloud takes care of all the infrastructure. You sign up, configure your bot (name, API keys, connected apps), and we deploy a dedicated instance for you on Fly.io. It runs 24/7 — no servers to manage, no Docker to wrestle with.
What you get:
- 🦞 Your own isolated AI assistant instance
- 💬 Connect WhatsApp, Telegram, Discord, Slack, Signal, iMessage
- ☁️ Always on — no need to keep your computer running
- 📱 Access and control from any device
- 🔄 Automatic updates and maintenance
- 🔒 Your own isolated environment — enterprise-grade security
The Tech Stack
This is a dev.to post, so let's talk tech:
- Elixir ~1.17 / OTP 27 with Phoenix ~1.8 and LiveView ~1.1 — the entire UI is server-rendered with real-time updates
- PostgreSQL via Ecto for data persistence
- Horde for distributed process supervision — each bot instance is a GenServer managed by a Horde DynamicSupervisor
- Fly.io for hosting — both the platform and user instances run on Fly
- Docker for local development, Fly Machines in production
- Stripe for subscriptions via their embedded pricing table
- Google OAuth + email/password for authentication
- Tailwind CSS + DaisyUI for the UI
- Bandit as the HTTP server
- Sentry for error monitoring
Architecture Highlight: Instance Management
Each user's bot is a separate process managed by a ContainerBackend behaviour:
# In development, instances run as Docker containers
Openclaw.Instances.ContainerBackend.Docker
# In production, they're Fly.io machines
Openclaw.Instances.ContainerBackend.Fly
The InstanceWorker GenServer manages each bot's lifecycle — creation, start, stop, restart — and reports status back to the LiveView dashboard in real-time via Horde's distributed registry.
Why Elixir?
Managing many concurrent, long-lived bot instances is exactly the kind of problem Elixir was built for. Each instance is a lightweight process, supervised and distributed across nodes. OTP's supervision trees mean that if a worker crashes, it gets restarted automatically. And LiveView gives me real-time UI updates without writing a single line of JavaScript.
Current Status
OpenClaw Cloud is in early access with a Hobby plan available. The onboarding flow walks you through:
- Creating your account
- Setting up your Discord bot token and API keys
- Configuring intents and behavior
- Deploying your instance
It's built on top of OpenClaw, an open-source personal AI assistant.
What's Next
- More chat platform integrations
- Voice features (TTS via ElevenLabs/OpenAI)
- Memory and context persistence
- Custom personality configurations
- Usage analytics dashboard
Try It Out
Early access is live at clawdcloud.net. I'd love feedback from the dev community — what features would make this useful for you?
If you're interested in the Elixir/Phoenix architecture, happy to go deeper on any part of the stack in future posts.
Built solo from Braga, Portugal 🇵🇹
Top comments (0)