PolyAI and Otter executives say voice AI has not had its ChatGPT moment yet
🎧 Listen to this story
Shawn Wen, PolyAI’s CTO, told TechCrunch that despite models able to speak and listen simultaneously, voice AI still needs faster reasoning to retrieve answers and sustain natural conversations. Alex Gay, CMO of Otter, said that identifying speakers, capturing intent and combining transcripts with organizational knowledge are important steps toward automating tasks and meetings. Otter is also working on “digital twins” that could represent people in meetings, but Gay said they would need to reproduce human expressions and relationships or amount to question-and-answer chatbots.
Wen and Gay said automatic speech recognition models often miss keywords. Gay said an inaccurate transcript undermines later summaries and actions, reducing trust in the platform. Both also argued that tools should tell people when they are speaking with AI or being recorded; Otter is considering notifying participants even when its bot is not present.
Why it matters · editorial interpretation
The comments suggest that adopting voice AI in meetings depends not only on more natural conversations, but also on reliable transcripts, integration with organizational knowledge, and transparency about recording and AI use. That puts accuracy and consent at the center of automating meeting tasks.