We are building a multimodal reasoning assistant. Especially focused on speech.