Show HN: Ori, an agentic runtime for infrastructure that must ask before it acts

Category: infrastructure

Tags: iot, agentic-ai, edge-computing, safety-critical, offline-first

Score: 6.3/10 (Innovation: 6, Technical: 6, Documentation: 7, Utility: 6)

Ori is an agentic IoT runtime that layers a tiered reasoning system (deterministic rules, local SLMs, optional gateway LLMs) with a Physical Actuation Trust framework to govern what autonomous agents may do in the physical world. It uniquely maps consequence levels to action tiers (A inform, B act within policy, C ask before acting, D protect immediately), enabling offline-first safety-critical automation on hardware as cheap as a Raspberry Pi.

Target audience: IoT engineers, embedded developers, edge AI practitioners, and infrastructure operators in low-connectivity environments

Repository: https://github.com/ori-platform/ori-runtime · Python · Apache-2.0 · 16 stars

View on Hacker News