TechCrunch reports that executives think voice AI is still waiting for a ChatGPT-like turning point.
TechCrunch shared a report on October 11, 2026, about executives who think voice AI has yet to reach its 'ChatGPT moment.'
11.3K
1 Source, first seen ago
Voice AI has progressed from basic transcription to systems that can listen and speak at the same time, but industry executives say that still falls short of a broader breakthrough. TechCrunch reports that PolyAI CTO Shawn Wen sees faster reasoning as the next hurdle: models need to retrieve answers quickly enough to keep a conversation feeling natural.
Wen said customer-service agents also need to sound less robotic and prove they can solve a caller’s problem. In his view, confidence can build over the first two or three exchanges, reducing the need to hand the call to a human.
Otter CMO Alex Gay pointed to a different stack of requirements for meetings: identifying speakers, capturing intent and connecting transcripts with organizational knowledge. Otter is also working on digital twins that could represent people in meetings, though Gay said those systems need enough emotional expression to support debate and strategic discussion rather than behaving like simple question-and-answer bots.
The executives also highlighted weaknesses earlier in the process. Wen said automatic speech recognition can miss important keywords and lose context. Gay said an inaccurate transcript can make summaries and follow-up actions wrong, eroding trust in the product. Both also emphasized disclosure: meeting participants should know when they are being recorded, and callers should know when they are speaking with AI.
—
Not ranked yet
—
Not ranked yet
