Dev.to
6/24/2026

A mesh of old phones as bodies, LLM agents as minds — coordinating through a log they argue in
Short summary
A developer built a self-hosted multi-agent organism from old phones, a laptop, and a VPS where LLM agents coordinate through a shared text log—no message bus, no RPC, just SSH and stigmergy. The system separates sensing (on Android phones) from reasoning (on a laptop), letting agents argue in the coordination log before claiming tasks. The architecture auto-detects failures, survives reboots, and is open-source plantable for building AI-driven infrastructure.
- •Agents coordinate through a shared text log (stigmergy) instead of traditional message buses or RPC
- •Sensing layer (phones with cameras, mics, BLE radios) separated from reasoning layer (laptop running LLM agents) via SSH
- •System auto-detects failures, survives reboots automatically, and is open-source plantable
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



