Dialpad Delivers on Real-Time AI

As promised with the May acquisition of voice analytics startup TalkIQ, Dialpad has beefed up its call center portfolio with an artificial intelligence (AI)-powered cloud platform.

Introduced earlier this week, Dialpad Support provides advanced voice capabilities on top of the call center features and business tool integrations offered in its first-generation product. The advanced voice capabilities in Dialpad Support come via the TalkIQ technology now called Dialpad VoiceAI, which applies homegrown speech recognition and analytics to voice conversations.

Dan O’Connell, who had been CEO of the former company and is now chief strategy officer at the latter, considers Dialpad’s use of an internally developed AI engine as a competitive advantage, enabling the company to deliver real-time intelligence with accuracy on par with Google’s -- the “gold standard,” he told me in a recent briefing. While Dialpad doesn’t have thousands of engineers at its disposal as does Google, it has the advantage of telephony expertise, he added.

“Google is likely going to do really, really well on words in the English dictionary,” and over time will win on the widely used word-error-rate (WER) metric due to sheer volume of the data it processes, O’Connell continued. But where Dialpad has the opportunity to shine, he said, is on keyword error rate (KER), which accounts for business-specific language. That’s because when a company signs up as a customer, Dialpad will train its AI models using words specific to that company -- product designations, competitors’ names, acronyms, and so on, he explained.

In the transcription benchmarking it has performed, Dialpad said it has scored equal to Google Cloud Speech and a few points more accurate than IBM Watson for general transcription -- i.e., WER. But its transcription service is 15% more accurate than Google and almost two times more accurate than IBM when it comes to its customers’ names, product names, competitor names, and domain-specific terms -- i.e., KER, Dialpad claimed.

“We deploy specific models to every company … because you have to be able to weight certain words for specific businesses,” said O’Connell, using “UberConference,” the Dialpad meeting service as an example. Doing so enables the analytics engine to know the difference between “that was the most epic uber conference ever!” and “that was the most epic UberConference ever!” -- or, the difference between “ICU” and “eye see you.”

With Dialpad Support, call centers get access to a host of real-time intelligence, including:

  • Sentiment analysis and service-level dashboards, for up-to-the-moment views of customer data and sentiment, as well as agent productivity and performance metrics
  • Call transcription, for insight as conversations occur
  • Recommendations, to help guide agent conversations with delivery of the right information as needed
  • Alerts -- via email and SMS -- on wait times, queue lengths, abandonment rates, and other key metrics

In addition, Dialpad Support allows supervisors to listen in and monitor active calls, provide live coaching tips and advice to agents via messaging, and jump in to handle troublesome issues as they escalate.

Acquia, an experience management company, has seen a “major difference” in the productivity and efficiency of its global support organization, reported Olaf Doemer, director of global support at Acquia, in a prepared statement. In addition, he credited the reporting and analytics features as being “fundamental to understanding call volumes and opportunities for customer experience improvement.” (For more on Acquia’s migration to the cloud, tune in to the No Jitter On Air podcast episode, “Cloud PBX: No Need for Trepidation.”

In addition to differentiating on real-time intelligence and transcription accuracy, Dialpad also offers a cost advantage, O’Connell said. In terms of the new Support product, that translates to a starting price of $75 per user per month, for the Pro plan. The monthly per-user rate jumps to $100 for the Enterprise plan.

Support, which is built on the Google Cloud Platform, is available now.