Jack Dorsey’s company, Block, has launched a new artificial intelligence voice agent called Dessa. What makes Dessa unique is that it was trained exclusively on real human conversations. Instead of mixing in computer-generated or scripted data, this AI model learned only from the ways people actually talk to one another.
This new approach aims to create an AI voice that feels more natural and human-like during interactions. In typical voice AI training, models often include various types of text and speech data, which can sometimes make conversations with these agents feel robotic or forced. Dessa’s focus on authentic human conversation could help reduce that feeling and lead to smoother, more engaging exchanges.
However, people who have interacted with Dessa report that conversations can sometimes feel awkward or uncanny. This means that while the AI sounds more natural in some ways, it still faces challenges in fully replicating human conversation. These quirks reflect the difficulty of teaching AI to understand and respond with the complex subtleties of real-life speech.
Why does this matter to sales teams, customer service agents, and small businesses? Voice AI agents are increasingly used to handle customer questions, book appointments, and assist with sales. An AI that can speak more like a real person could improve customer satisfaction by making automated interactions feel less mechanical and more genuine.
For small businesses without large customer service departments, Dessa-style voice agents could offer a way to provide friendly, human-sounding help at any time of day. This might make customers more likely to stay engaged and follow through with purchases or support requests.
While Dessa shows promise, it is still early days for this kind of conversational training on AI voice agents. Organizations considering voice technology should watch closely to see how models like Dessa evolve, especially if they aim to combine natural speech with reliability and understanding.
The launch of Dessa highlights a broader shift in AI development: focusing not just on accurate speech but on authentic conversation patterns. This could inspire future AI voice tools that better match real human interaction, potentially transforming how businesses communicate with customers by voice.
Readers interested in voice AI should keep an eye on updates from Block and other companies experimenting with conversational training methods. As AI continues to learn from our natural ways of talking, the tools for sales and service could become more helpful and less frustrating for everyone involved.