Voice AI is attracting investment and new products, from customer-service agents to meeting assistants. Yet executives at PolyAI and Otter say the technology has not reached the kind of mainstream breakthrough ChatGPT achieved.
PolyAI CTO Shawn Wen says systems can now listen and speak at the same time, but they still need to reason and retrieve information faster to make conversations feel natural. In customer service, he adds, people are more likely to rely on an AI agent when it can clearly understand them and resolve their issue.
Otter CMO Alex Gay points to another opportunity: combining speaker recognition and intent detection with an organization's knowledge. Such capabilities could support meeting automation and digital representations of participants. For these tools to feel useful in collaborative settings, however, they need to convey the nuance and emotional expression that make human discussions engaging.
Accuracy and transparency remain essential
Both executives highlight transcription accuracy as a foundation for dependable voice AI. If speech recognition misses a key word, the resulting summary or follow-up action may also be incorrect, undermining confidence in the system. Language understanding is another area where continued progress is needed.
Clear disclosure is equally important. People should know when they are interacting with an AI or when a meeting is being recorded. PolyAI and Otter see transparency as part of building trust in voice-powered tools.
As reasoning, accuracy and disclosure improve, voice AI could become a more capable and trusted layer for everyday communication and work.
