An AI research company building open multimodal foundation models and efficient architectures, spanning language, audio, and neural signal decoding.