Netvora logo
Submit Startup Subscribe
Home About Contact Submit Startup Subscribe

Phonely’s new AI agents hit 99% accuracy—and customers can’t tell they’re not human

Comment

Phonely’s new AI agents hit 99% accuracy—and customers can’t tell they’re not human

Phonely’s new AI agents hit 99% accuracy—and customers can’t tell they’re not human

AI Phone Support Breakthrough: 70% Faster Response Times and 99.2% Accuracy

By Netvora Tech News


A groundbreaking collaboration between Phonely, Maitai, and Groq has overcome a significant hurdle in conversational artificial intelligence: the annoying delays that immediately give away the fact that callers are talking to a machine. This achievement could revolutionize the way customer service operates, as Phonely's customers are already replacing human agents with AI-powered support. Phonely, a leading provider of AI phone support, has reduced response times by more than 70% while boosting accuracy from 81.5% to 99.2% across four model iterations. This surpasses the 94.7% benchmark set by GPT-4o by 4.5 percentage points. The key to this success lies in Groq's new capability to instantly switch between multiple specialized AI models without adding latency, orchestrated through Maitai's optimization platform. The elimination of delays solves the "uncanny valley" of voice AI, a subtle phenomenon that makes automated conversations feel distinctly non-human. For call centers and customer service operations, this breakthrough could be transformative. One of Phonely's customers is replacing 350 human agents this month alone. Traditional large language models, like OpenAI's GPT-4o, have long struggled with what appears to be a simple challenge: responding quickly enough to maintain natural conversation flow. While a few seconds of delay barely registers in text-based interactions, the same pause feels interminable during live phone conversations.

Why AI Phone Calls Still Sound Robotic: The Four-Second Problem

The issue lies in the fundamental architecture of large language models. They are designed to process text input, not audio conversations. This leads to a delay of around four seconds, which is sufficient to make the conversation feel robotic. To overcome this challenge, AI systems need to be able to process and respond to audio input in real-time, without sacrificing accuracy.

How Three Startups Solved AI's Biggest Conversational Challenge

  • Maitai's optimization platform enabled the instantaneous switching between multiple specialized AI models, reducing latency and improving accuracy.
  • Groq's new capability to switch between AI models without adding latency was the key to achieving faster response times and higher accuracy.
  • Phonely's expertise in AI phone support and customer service operations helped to identify the specific challenges and develop solutions to overcome them.
This collaboration demonstrates the potential for innovative solutions to overcome the limitations of conversational AI. As the technology continues to evolve, we can expect to see even more significant improvements in the way AI-powered customer service interacts with human customers.

Comments (0)

Leave a comment

Back to homepage