Building Ori, an open-source agentic IoT runtime with a Physical Actuation Trust framework.

The core idea: every AI agent acting in the physical world must formally earn the authority to act, tier by tier, from informational alerts through to safety-critical relay control. Runs offline on a $55 Pi.

First deployments are underway in Lagos.

Happy to answer questions about the safety architecture or the offline reasoning approach.

[dead]