Nuance Communications is trying to make it easier to program a convincing Virtual Assistant. The company revealed its latest innovation at its recent Customer Experience Summit, explaining that Project Pathfinder uses machine learning to improve the conversational capabilities of chatbots and VAs.
According to Nuance, Project Pathfinder examines existing chat logs and call transcripts to build dialog models based on real human conversations. That drastically reduces the cost of developing a Virtual Assistant and mitigates the amount of human error that creeps into the software. The end result is a specialized Virtual Assistant that can have a contextual two-way conversation about subjects ranging from healthcare to financial services.
“While today’s virtual assistants can reliably facilitate basic gateway questions and answers—such as ‘What’s my account balance?’—most are not able to have a real conversation or address more complex questions,” said Nuance EVP and GM Robert Weideman. “Unlike other conversational AI tools—which are predominantly programmed by humans—Project Pathfinder automates the build for human-to-machine interactions thanks to our advancements in natural language understanding (NLU) and intent discovery.”
Nuance has been making major moves since selling off its document imaging division to focus more on AI and cloud solutions late last year, forging partnerships with MEDITECH and Allied Irish Banks while opening a new research lab in Germany. Project Pathfinder will widely be available in the summer of 2019, and should further increase the appeal of the company’s voice recognition solutions.