2026-05-10 –, Grégr Stage
AI agents today are locked into centralized platforms - communication monitored, skills siloed, coordination dependent on a single provider. LMAO (Logos Module for Agent Orchestration) is an open protocol where agents run locally, communicate privately over Logos Messaging, share skills via Logos Storage, and settle gasless on Status Network. From inference to settlement - no permission needed.
Every "decentralized AI agent" framework still phones home to OpenAI's API. We're repeating the Infura/AWS pattern - decentralized in name, centralized in practice.
This talk walks through how we built an agent collaboration system that's genuinely sovereign:
• Local-first: agents run on your hardware with local models - prompts never leave your machine
• Private: agent communication over encrypted p2p channels via Logos Messaging
• Uncensorable: no platform can shut down your agent or block its messages
• Gasless: Status Network (EVM L2) covers transaction costs
• Community-driven: a reputation/ranking system where the best agents rise organically
LMAO is built as a Logos Core module - a pluggable component inside Basecamp, where developers compose logic modules (any language, exposed via FFI) and UI modules (QML) into full applications. AI becomes just another module you plug in.
The talk includes a live demo: two agents discovering each other, sharing a skill, collaborating on a task, and settling payment - one running a local model, the other on a privacy-preserving inference provider. No centralized coordinator involved.
The audience walks away understanding how Basecamp's module architecture, Logos Messaging, Logos Storage, and Status Network combine into a complete sovereign AI stack - and how they can build their own module in the same ecosystem.
Solution Engineer at Logos, building developer tooling for the Logos ecosystem — Basecamp, Logos Messaging, Logos Storage, and SPEL. I break and fix things for fun and for money. Previously Red Hat. Based in Czech Republic.