Launched this week
PrivateClaw
First AI agent platform with provable E2E encryption.
20 followers
First AI agent platform with provable E2E encryption.
20 followers
Your OpenClaw host can read every conversation you have with your AI agent. We built infrastructure to change that. Introducing PrivateClaw, the first AI agent hosting platform with provable e2e encryption. - OpenClaw running in a Trusted Execution Environment (TEE) - Inference running in a TEE - Together: Verifiably E2E private agentic flow Getting started takes one command in your terminal: `ssh privateclaw.dev`
jared.so
TEE-backed agents with one-line ssh onboarding is a sharp wedge here. How do you handle attestation drift when AMD pushes microcode updates mid-session?
@mcarmonas The privateclaw verify CLI that all users can run on their PrivateClaw CVMs (confidential virtual machines) runs in real time against AMD's live chain, so a TCB bump just means the next verification fetches a new VCEK cert and validates it normally.
CVMs can't be patched in place because microcode updates require a host reboot, which ends the session, and the reconnect would trigger a fresh attestation on the CVM anyways.
The `ssh privateclaw.dev` onboarding is the cleanest I've seen for a privacy based product. Just directly shell into a verified CVM with your public key, and that's it. Super cool stuff!
this is awesome. SSH sign up super slick
huge difference: can freely, confidentialy enter personal and business info. and it stays personal. like contracts. and code.