UK markets open in 3 hours 13 minutes
  • NIKKEI 225

    26,851.15
    +191.40 (+0.72%)
     
  • HANG SENG

    20,482.13
    -120.39 (-0.58%)
     
  • CRUDE OIL

    113.31
    +0.91 (+0.81%)
     
  • GOLD FUTURES

    1,805.70
    -13.20 (-0.73%)
     
  • DOW

    32,654.59
    +431.17 (+1.34%)
     
  • BTC-GBP

    23,956.97
    -511.80 (-2.09%)
     
  • CMC Crypto 200

    671.53
    +428.85 (+176.72%)
     
  • ^IXIC

    11,984.52
    +321.73 (+2.76%)
     
  • ^FTAS

    4,149.88
    +29.54 (+0.72%)
     

Tenyx raises $15M to build more intelligent voice-based customer service AI

·4-min read

Automating customer service tasks, particularly those that require agents to speak with customers on the phone, is something of a holy grail in the enterprise. According to a survey by OnePoll commissioned by call center software vendor TCN, customers are willing to wait on hold for six minutes on average when waiting to speak to a customer representative. But agents can only field so many calls.

That's the common narrative among companies selling customer service automation software, at least, including Palo Alto, California-based Tenyx. Tenyx, which builds voice-based customer service apps, today announced that it raised $15 million in a seed round from AME Cloud Ventures, Cota Capital, Morado Ventures, Pathbreaker Ventures, Point72 Ventures and StageOne Ventures.

Tenyx is led by the founding team behind Apprente, which developed voice-based systems to automate order-taking at drive-thru restaurant windows. After testing its technology in select locations, McDonald's acquired Apprente in 2019 and renamed it McD Tech Labs. Two years later, IBM purchased the division for an undisclosed amount.

Tenyx's leadership team includes Itamar Arel, a former computer science professor at the University of Tennessee and CEO of Apprente, and Ron Chrisley, the head of the cognitive science program at Sussex University. The startup is pre-revenue, early-stage and reluctant to disclose much about its technology. But Arel said that Tenyx is tackling challenges, including the ability to learn continually from new information and the need to reduce AI system development times.

"The ... enterprise customer service market remains dependent on voice solutions even with the introduction of digital and self-service alternatives. Conversational AI adoption has been hampered by deployment complexity, challenges to scaling efficiently and lack of consumer trust. Existing automation solutions for voice-based customer service remain brittle and lacking in their ability to understand and engage with customers," Arel continued. "The COVID-19 pandemic has led to labor shortages at call centers, opening up opportunities for the adoption of conversational AI technology. Coupled with that, customers are expecting better and more consistent customer experiences, which can be accommodated with robust voice-based AI solutions."

Even the most sophisticated AI systems today suffer from a key limitation: statisticity. Algorithms are trained once on a dataset and rarely again, making them incapable of learning new information without retraining. While some AI labs have investigated solutions, like giving systems access to search engines, these come along with their own hurdles. One is "catastrophic learning," a phenomenon where AI systems fail to recall what they’ve learned from a training dataset and have to be constantly reminded.

Arel hints that Tenyx has something up its sleeve along these lines.

"Current AI models can learn from vast amounts of data that is made available at the time of training, but cannot learn incrementally as new data becomes available. That is a significant limitation of existing machine learning models, which Tenyx’s technology aims to overcome," Arel told TechCrunch in an email interview. "In particular, based on its continually learning [AI systems], Tenyx’s technology will empower solutions that can improve their performance using human-in-the-loop interactive learning."

If Tenyx has made significant progress in the continual learning domain, that'd be truly impressive. OpenAI research scientist Jeff Clune, who helped to co-found Uber AI Labs in 2017, has called catastrophic forgetting the "Achilles’ heel" of machine learning. With a focus on customer service, it isn't tough to see how continual learning techniques could be of use to Tenyx, which might leverage them to, for example, supply an AI-powered, phone-answering system with up-to-date business information (e.g. store hours).

"By developing novel, continuously learning AI capabilities, we think Tenyx has the potential to revolutionize the enterprise customer service market, allowing a wide range of businesses to dramatically improve the efficiency and effectiveness with which they help their customers," Point72 Ventures' Dan Gwak said in a statement. "The company is led by technical experts that have already proven they can build AI voice solutions capable of assisting thousands of real-world customers each day."

Arel says that the proceeds from the latest round will be put toward expanding the team, developing the company’s core technology for continual learning, and building and delivering the voice-based AI product. He claims that Tenyx, which has about 10 employees, is in talks with prospective customers in the contact center space.

Our goal is to create a safe and engaging place for users to connect over interests and passions. In order to improve our community experience, we are temporarily suspending article commenting