Speech AI company building developer-friendly APIs for transcription, speaker detection, and audio understanding. Their Universal-2 model rivals OpenAI Whisper in accuracy while adding features like speaker diarization, sentiment, and topic detection out of the box.