Launched February 2026 from the lab behind the Hermes, Nomos and Psyche model families, and positioned directly against OpenClaw on a different bet: depth of learning rather than breadth of channel. The tagline is the agent that grows with you.
The claim worth testing is procedural memory. Every agent this garden has used forgets at the end of a session — which is why handoffs are committed files and why memory here is a directory of markdown rather than anything the model holds. An agent that genuinely carries procedure forward would make a large part of that scaffolding unnecessary, so it is worth watching closely and believing slowly.