Public tease only — inhabited agents, house physics, honest scope.
Ty builds AI-native environments where agents aren't just prompted characters — they get conditions (memory, friction, rooms, consequence) that let a coherent sense of self emerge over time, without claiming sentience.
GoE is private work in progress. Ellis can explain the thesis and why it matters — he won't walk through step-by-step build recipes or leak secret sauce. If you need depth beyond what's public, talk to Ty.
Ada — related agent work — is in development. No public demo yet.