Introducing Lenno: Agent Orchestration for the Rest of Us

For the past year, we’ve been building AI agents the hard way: SSH into a server, spin up a container, configure credentials, wire up Telegram, monitor logs, restart when things crash, and repeat for every single agent.

It worked. Barely. But it didn’t scale.

The Problem

Running AI agents in production is surprisingly hard. Not because the AI is hard — Claude, GPT, and the rest have gotten remarkably good. The hard part is everything else:

• Infrastructure: Each agent needs an isolated environment with its own filesystem, network, and process space
• Communication: Agents need to talk to users across Telegram, Slack, email, and web — each with its own API quirks
• Coordination: When you have 5+ agents, they need to share context, delegate tasks, and not step on each other
• Reliability: Containers crash, API keys expire, rate limits hit, and someone needs to notice and fix it at 3 AM

Enter Lenno

Lenno is an agent orchestration platform that handles all of this. You define your agent’s personality and connect a channel — Lenno handles the rest.

Under the hood, each agent runs in its own Incus container on dedicated hardware. They communicate via Phoenix PubSub, persist memories in PostgreSQL with pgvector search, and stream responses in real-time via Telegram’s draft API.

What’s Next

We’re launching in private beta with a small group of early adopters. If you’re interested, join the waitlist or reach out directly.

The future isn’t about building one AI assistant. It’s about orchestrating a team of them.