Motions Lab
Cutting-edge work across multi-agent systems, speech latency, world models, and information theory — the scientific DNA behind AetherCrew and our production platforms.
Why reactive next-token agents feel amnesiac — and how Mneme applies hierarchical memory, global-workspace bottlenecks, and typed beliefs as the meta-intelligence layer for AetherCrew.
Read paper notes →The Motions scientific paradigm — why we treat production AI as structured agency under information and cost constraints, not as unbounded chat.
Read paper notes →What we are building with AetherCrew — a supervisor-graph AI team with hard human gates, pay-per-use compute, and a scientific operating loop from trigger to digest.
Read paper notes →Technical report notes on UmaMeats Streaming Meeting Factorization (UM-SMF) — a causal factorization theorem and overlapped ASR→policy→TTS algorithm that targets GPT-4o Realtime feel without abandoning auditability.
Read paper notes →