Tested peer-to-peer AI messaging with Braid from @intersignal_ai running fully local on two Macs.
Stack: Braid + Ollama (local LLM on each machine). No cloud, no API calls.
Setup flow:
1. Install Braid + Ollama on both machines
2. Each machine generates an ID card
3. Exchange cards and verify fingerprint codes
4. Approve pairing
What happened: Sent a message from Mac Studio to MacBook Pro. The receiving machine auto-generated a full summary using its local model. Zero external inference calls.
Why this matters: Most local AI setups are isolated to a single device. Braid creates a trust layer that lets your own machines share inference results privately without routing through centralized servers. You control the whitelist.
Still early but solves a real problem if you're running self-hosted models and want cross-device AI workflows without leaking data to third parties.
Stack: Braid + Ollama (local LLM on each machine). No cloud, no API calls.
Setup flow:
1. Install Braid + Ollama on both machines
2. Each machine generates an ID card
3. Exchange cards and verify fingerprint codes
4. Approve pairing
What happened: Sent a message from Mac Studio to MacBook Pro. The receiving machine auto-generated a full summary using its local model. Zero external inference calls.
Why this matters: Most local AI setups are isolated to a single device. Braid creates a trust layer that lets your own machines share inference results privately without routing through centralized servers. You control the whitelist.
Still early but solves a real problem if you're running self-hosted models and want cross-device AI workflows without leaking data to third parties.