Core Rth - The AI agent platform that asks before it acts.

Core Rth is a governed AI assistant you run locally. Like OpenClaw, it connects to WhatsApp, Telegram, Email, and Discord — but every action the AI proposes must survive a governance layer before execution. Built for engineers who automate physical systems (IoT, robotics, drones) and can't afford hallucinating agents with root access.

Hey Product Hunt! 👋 I'm Christian, and I built Core Rth because I got tired of watching AI agents run commands on my systems without asking. If you've used OpenClaw or similar AI assistants, you know the magic — talk to an AI from WhatsApp, have it automate your life. Core Rth does all of that. Same channels, same local-first philosophy. The difference is that I come from 20 years of enterprise cybersecurity. So I couldn't ship a product where an LLM has unrestricted access to the filesystem, the network, and — in my case — actual robots and drones. Every action in Core Rth is a governed proposal. The AI suggests, the Guardian evaluates, and you decide. If something goes wrong with physical hardware, there's a big red E-Stop in the dashboard. I'd love feedback on: 1. The security model — is The Guardian's proposal-first architecture practical for your workflows? 2. The Model Router — how do you currently handle multi-model routing? 3. The Reality Bridges — anyone else connecting AI to physical systems? GitHub: https://github.com/rthgit/CORE-RTH Happy to answer any questions!

Core Rth - The AI agent platform that asks before it acts.

Replies