Blog
Notes from the lab.
Organismic AI: Why True Intelligence Must Be Grown, Not Pretrained
Announcing Organismic Artificial Intelligence (OAI) and BIB 2: a 540 Hz synthetic nervous system that learns continuous sensorimotor control on a single CPU without backpropagation, cloud servers, or frozen weights.
Read →ATOMIC: Breaking the CPU Memory Wall with Zero KV-Cache Linear Recurrence
Why Transformers stall on consumer laptops, how Gated Linear Recurrent Attention (GLRA) maintains an invariant 192KB state, and how we trained a Chain-of-Thought reasoner on a single CPU in 12.6 minutes.
Read →Project Cambrian: Bridging 30 Years of Artificial Life with First-Principles Synthetic Biology
Announcing Project Cambrian (v1.0.0). How we unified 30 years of Artificial Life lineage—from Karl Sims supercomputer creatures to real-time 8-organ digital organisms executing at 60 FPS on consumer CPUs.
Read →Project Big Bang: The Journey to Biologically Inspired Brains
The full story of BIB and Project Big Bang—how we built a virtual evolutionary environment to let neural networks survive, feed, and grow, and why biological fidelity beats standard deep learning.
Read →The Great Bifurcation
How the BIM 4 prototype proved so successful that the project needed to split to honor both the commercial NLP disruptor (BILM) and the synthetic lifeform (BIO).
Read →MAUL: The Hypothesis
Why specialization beats generalization. How routing requests through dedicated agents — math, code, search — outperforms a single model doing everything.
Read →BIM: The Full Hypothesis
Transformers are frozen. BIM is not. The case for a biologically-grounded neural substrate that learns continuously, without catastrophic forgetting, on a CPU.
Read →