Voice Assistant, can you say ‘Kismet’?
Either that or part of a grand plan, because ChatGPT is lighting a rocket under Nuance’s prominence in the conversational AI underlying contact centres, chatbots and interactive voice response (IVR) domain.
Planned to unplanned, Microsoft’s later deal, announced in January this year, to take a $10 billion stake in OpenAI augments its Nuance acquisition tidily. (Nuance has 75% of the Fortune 100 as customers and a deep footprint across government, financial services, healthcare, retail and telcos, among other verticals.)
Over in the UK recently for a user conference, Tony Lorentzen, SVP of Intelligent Engagement at Nuance and a twenty-something-year veteran there, says that its domain has already “dramatically changed” over the last three to five years as companies have left the physical world and moved to digital channels for sales and service.
With lockdown having already driven a pivot to more chat/IVR usage, the Redmond mothership factor and a new ChatGPT-powered Nuance product line called Conversation Boosters, things are about to get even more interesting.
As Nuance notes, while most organisations' chatbots can "automate frequent, transactional tasks, they’re not suited for answering a broader set of customer questions, including complex questions that require a level of reasoning and summarization."
Guess what is? It starts with "GPT" and ends with an apparently rapidly escalating number set -- and improving LLM performance.
The readymade ecosystem could be key to its success.
"What Microsoft and Google are doing to build ChatGPT into their platforms is a good guide to what needs to happen - see ChatGPT as a capability within the bigger platform," says Jon Collins of analyst house GigaOm. "You can't just deploy a chatbot, for example; it needs to be integrated with your back-end systems and practices."
Nuance: ChatGPT evolution… or revolution?
“For us, ChatGPT is an extension of all the AI we’ve been doing,” Lorentzen says. “[It’s] the biggest thing for us right now [for customers that want to] do more with less. [They’re looking at] how you make agents more efficient and how to automate more, and now with ChatGPT you can apply that to the long tail.”
Nuance has been “toying” with the tech since December “and seeing how you apply it in the enterprise virtual assistant space,” Lorentzen says. “Being part of Microsoft, we’re obviously part of a large AI domain and we were exposed to that technology. It’s a game-changing technology that’s here to stay. Now [companies can] find even more ways to reply [to customers] and it gives them an enormous amount of optionality.
“I’ve worked with many contact centres and there’s always been a question of what do you automate and why now. With ChatGPT, it eliminates a lot of that complexity for faster time to value and faster time to market.”
That means IT and customer service leaders get to focus on high-complexity interactions and go a lot further without calling Nuance or other vendors/SIs, he adds.
Of course, it’s not (and never is) a universal panacea though.
“If you think about virtual assistants, many companies have built the very specific neural network capabilities we have today and those are finely tuned for their use cases. It’s an extension, not a replacement technology,” Lorentzen cautions. However, it will lead to more sophisticated “extended enterprise conversations”.
Giving new voice
Nuance’s roots are in voice commands, dictation and speech recognition: it owns the Dragon speech recognition suite and provided the original back-end NLP algorithm for Apple’s Siri. And voice has of course been closely watched (or listened to) with the rise of consumer voice assistants but Lorentzen says voice still has obvious limitations such as being understood noisy environments or with complex instructions: decoding a long alphanumeric string on a shipping order, for example. Lorentzen doesn’t expect consumer voice assistants to change that soon.
“[Consumer] voice assistants are simple assistants that do FAQs; anyone can put it in the system and create a bot out of it… that’s table stakes, a commodity. What’s going to differentiate is creating highly conversational experiences and being able to transact. That’s what’s going to set Nuance apart. Experience will differentiate it. Voice has become an escalation channel. When you decide to interface with a brand, it’s usually a digital channel and voice comes in when you need to talk to someone. Voice has got to be a lot better than five to 20 years ago and has to be a highly conversional experience.”
Could advances in voice lead to some level of disintermediation where consumers interact with brands directly over smart speakers and similar?
“I don’t believe so quite honestly,” Lorentzen says. “To make voice systems meaningful [at an enterprise level], you need access to CRM systems, billing systems, be able to do a lot of integrations. Will Alexa or Google Assistant do that? No, I don’t think so.”
But what about Amazon? By tying leadership in online retail to leaderships in voice assistants, doesn’t it have the chance to make buying by voice a commonplace activity as it (sort of) is with its groceries ordering?
“Could be, but there’s an element of trust there too,” Lorentzen counters.
Comply or die
And what of compliance: is Nuance concerned over data sovereignty and the sand-shifting world of regulations and data privacy/protection?
“They’re there to protect consumers and we spend lots of time and energy trying to protect those areas,” he says. “We believe we’re stewards of customer data. It’s a very, very important area. You have to think twice, let’s put it that way [and] you can’t leave that stuff out for interpretation.”
As for the Microsoft relationship, Lorentzen says learning “how to deploy at the Microsoft scale has been eye opening” after a long history as a leader of an (albeit sizeable) niche.
And with ChatGPT on board and other aforementioned winds blowing in its favour, Microsoft and Nuance have a chance to take the conversation about customer interactions even further and deeper.