Voice AI has drawn billions in investment and a steady stream of new models, but PolyAI CTO Shawn Wen says the technology has not yet had its "ChatGPT moment." Speaking at the HumanX conference, Wen acknowledged that full-duplex models can now speak while listening, but argued the next challenge is making reasoning fast enough for models to fetch answers quickly and keep conversations feeling natural. He also said customer-service agents should not sound robotic and need to give callers confidence that they can solve problems.
Otter CMO Alex Gay takes a different angle, emphasizing speaker identification, intent capture, and tying that to organizational knowledge as key steps toward automation. Otter is working on digital twins that might represent people in meetings, but Gay said the output voice must carry the same emotive expressions as a human conversation. Without that, he argued, an avatar is just a Q&A chatbot.
The two executives agree on the importance of transcription accuracy and transparency. Wen said automatic speech recognition models often miss important keywords, which undermines context. Gay echoed that, noting that flawed transcription makes all downstream actions flawed and erodes trust. On transparency, Otter wants to notify participants when meetings are recorded, and PolyAI says enterprise calls should make it clear that callers are talking to an AI.