To paraphrase Isaiah 11:6, sometimes “a little start will lead them.” That’s the case when Minerva CQ, a start-up specializing in “real-time collaborative intelligence for the enterprise,” announced that a major player in the world of EV Mobility (think of ‘charging’) implements its solution to help customer support agents quickly understand and respond to customer comments and questions. It represents the first commercial deployment of Riva Enterprise from NVIDIAa package of the GPU Giant’s Automated Speech Recognition (ASR), Text-to-Speech (TTS), and Conversational AI resources.
The deal marks a coming-out party for Minerva CQ, whose software platform listens as customers use their own words to tell a voice-activated interactive speech response (IVR) about the purpose of their call. It then uses that information to populate an agent screen with suggested wording or actions to resolve the customer’s problem when the call is transferred to a live agent. At that point, the system takes an approach that CMO Daniel Hong calls “Waze in the customer service space.” It stays on the line and listens to every word and provides further suggestions based on natural language understanding and sentiment analysis.
At the end of the call, Minerva’s proprietary Natural Language Generation (NLG) resources create a conversation summary that can be written to systems of record, including Salesforce, Zendesk, Servicenow, Microsoft, and Oracle. The summary can be integrated into quality assurance and agent training programs and can also speed up post-call workflows by recognizing specific triggers for follow-up. This approach supports coaching agents during the call and in post-call actions that accelerate the process of resolving recognized issues.
Validation for NVIDIA’s approach to conversational AI
NVIDIA has benefited from the explosive growth in Enterprise AI since software developers determined that the Graphics Processing Units (GPUs) it developed for gaming and the computer-generated imaging (CGI) that permeate today’s blockbuster movies were also well-suited for machine use. learning, speech processing and natural language processing. In May, it rolled out Riva 2.0, calling it a “world-class speech AI SDK” that provides developers with tools to customize voicebots and speech-based applications that take advantage of highly accurate speech recognition in seven languages and text-to-speech rendering of both male and female voices.
Riva’s product managers focused on internally developed deep learning and neural models to support ‘generic’ applications. They encouraged ISV (Independent Software Vendors) to pursue vertical opportunities through their TAO Toolkit to create custom implementations. Minerva is a customer of NVIDIA Riva Enterprise is a package offering that includes Riva, TAO and enterprise support.
Fast, accurate NLP and NLG achieving commodity status
Expect to see many resources with the ability to capture 100% of conversations between companies and their customers. Minerva CQ will increase its presence in the emerging green energy field and find agreements with a wider community of utilities and telecommunications companies. Expectations among developers and customer experience professionals are rising as Minerva CQ, NVIDIA and their Conversational AI cohort demonstrate how easy it is to integrate their technologies into customer and agent workflows and drive positive business outcomes.
Categories: Intelligent assistants