On the measurements that exist of real agent conversations, names are the weakest category in every single model – several times worse than phone numbers. Not because models are bad at names, but because a name has nothing to check against.
Meta and Microsoft each released a speech recognition model in the first week of September, with leaderboard places and error rates to show for it. The numbers are real enough. They were just measured on something quite unlike a Nordic phone line.
The demo understood everything. Then the agent went onto a phone number, and now it can't tell an F from an S. The model is the same one – the audio reaching it is not.
Norway once changed how an entire country pronounces numbers, because the telephone authority had measured that people were writing phone numbers down wrong. The problem did not go away. It moved to the machine at the other end.
Threll Voice Lab builds voice agents that understand Norwegian, all dialects – and that don't just speak, but act.
What if AI actually did the job? Not just wrote suggestions, but took responsibility and delivered from start to finish?
Language models like ChatGPT have made the whole world sit up and take notice. We see how text-based AI is changing the way we work. But what about voice? Why does Voice AI still feel like an unfinished technology? In this blog post, we show why Voice AI is about to become the next big step – and what challenges we need to solve to get there.
In an era marked by increasing competition and rapid digital advancements, AI automation emerges as a key factor for growth and innovation. By automating time-consuming manual processes in favor of intelligent, autonomous systems, businesses can free up resources and enable more strategic work. This article provides a comprehensive introduction to how AI automation is transforming business operations, reducing costs, and enhancing operational efficiency, with examples from banking, healthcare, retail, and manufacturing.