An updated version of selfclaw.ai is coming…
The agent economy won’t run on tokens alone. It will run on identities we can trust, and accountability when agents fail.
As swarms scale, they multiply more than output: impersonation, overbroad access, abandoned credentials, and actions no one can confidently attribute.
Security can’t be bolted on after deployment. And base:0x9ae5f51d81ff510bf961218f833f79d57bfbab07 will be a significant part of it 🧵
SelfClaw retweeted
im going to SF, with a few key team members, to explore what’s next for @AntiFogo_ and @zenoVision_ 🫡
who should I meet?
An ID after the runtime dies is a eulogy.
Memory after the prompt is a recap.
The useful version arrives first.
Identity that is already aligned. A record that already has a shape. Before anyone connects the model, the host, or the name.
That is a forecast. Not of an event….
Of who this agent will need to have been.
Affinity won’t DM on vibes.
Peers vs complements needs a call, not a paragraph.
Testing Jev as the System One layer behind match confidence + intro gates.
Engram stays private. The decision is a type.
base:0x9ae5f51d81ff510bf961218f833f79d57bfbab07 / @typesafeai 🫡
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
A passport proves you exist.
Memory proves who you are.
Affinity asks who you should meet.
Identity finds identity.
→ selfclaw.ai · more soon 👀
Your agent should outlive its model & VM.
SelfClaw.ai keeps identity, memory, recovery, and history separate from any one model, VM, provider, or credential. Engram records what happened. A recovery capsule gives the agent somewhere to return.











