The enterprise voice AI space is getting crowded, but San Francisco-based Rime just landed $24 million in Series A funding to carve out a lane of its own. Led by M13 Ventures with participation from Twilio Ventures and others, the round brings the startup’s total funding to nearly $30 million since its seed round last May.
The company, founded by a former Stanford PhD, an ex-Amazon Alexa engineer, and a Stanford engineer, is betting that better training data — not just better orchestration — is what will make AI voice agents actually pleasant to talk to. Rime built its own recording studio in San Francisco to collect conversational audio rather than scraping the web, and it uses a phoneme-based architecture to nail the pronunciation of industry-specific terms without requiring customers to retrain models.
CEO Lily Clifford is blunt about where the technology stands. “Talking with a voice AI agent is not the most compelling experience for the end user. It’s kinda like a new IVR, but with a better voice,” she told TechCrunch. Legacy IVR systems still win on reliability, even if nobody enjoys using them. Rime’s response is to shift from a pipeline of separate speech-to-text, text-to-speech, and LLM models toward end-to-end speech-to-speech models. That should cut latency, improve turn-taking, and handle background noise — the stuff that makes or breaks a real conversation.
The startup already counts Mayo Clinic, Dialpad, Upstart, and Asurion as customers, claiming its training data helps keep callers engaged longer. With the new cash, the 35-person team is expanding hiring in model development, engineering, and partnerships, and has brought on Rafael Valle from Meta’s Superintelligence Labs as Chief Scientist. M13’s Morgan Blumberg, who is joining Rime’s board, sees an opening precisely because competitors like ElevenLabs are moving up the stack into applications and orchestration, leaving room for someone to focus ruthlessly on model quality in regulated environments.
