Conversica CEO defines what ‘enterprise-ready ChatGPT’ means for CX
If brands are excited but nervous about the prospect of enhancing the customer experience they deliver with generative artificial intelligence technologies like ChatGPT, Jim Kaskade has some simple advice for them: “You’ve got to filter out the sex, drugs and rock n’ roll.”
The CEO of Foster City, Calif.-based Conversica Inc. wasn’t trying to play down recent stories of ChatGPT making inappropriate or offensive comments. He simply suggested that using any generative Ai technology for business will naturally require fine-tuning it to act more professionally – almost as though it had been hired rather than deployed.
“If you’re Nike, you want feel like their brand gets discussed (by ChatGPT) as if it was an employee of Nike, not Adidas,” Kaskade told 360 Magazine. “If you lead with a brand-specific language model, you’ll get 99 per cent of the answers right.”
Developing that kind of language model is part of the value add Kaskade’s company believes will make Conversica Chat, which was released earlier this week, the first truly enterprise-grade ChatGPT solution of its kind. Beyond that, he said the company has also scoured existing language models from Google and elsewhere to curate the broadest and most appropriate kind of “small talk” that could come up.
This means Conversica Chat will redirect and avoid discrimination constructs such as sexism and hate speech. It will also support more than 100 languages and engages users beyond basic Q&A to drive specific business outcomes, such as lead capture and conversion.
Though generative AI in general and ChatGPT specifically have attracted a flurry of interest in the last few months – no-code application provider Folloze also launched a ChatGPT tool for businesses this week – Kaskade said Conversica has been actively incorporating the technology into its applications since 2020. He said this was given by the fact that for many brands, the experience of using chat as part of their customer experience has been underwhelming.
“I think chat, as a channel for marketeers, has been kind of a checklist item. When you’re trying to drive demand gen for your sales teams, chats is just one of many items,” he said. “Even more than lead gen, it can help somebody throughout the customer lifecycle – all the way to advocacy.”
The first generation of chat, for instance, has often been seen as a mechanism to deflect customer support calls that would have taken up agent’s time in the contact center. Since then, chat has been recognized for its potential to drive sales, but with limited success, Kaskade said. Part of the issue has been recognizing that chat might represent only one stage of the customer journey.
“B2B buyers, for example, don’t want to be on chat forever – they’ll go to e-mail,” he pointed out.
“Some people like to talk via SMS, some people WhatsApp, some people on Twitter, some people still actually want that old fashioned ability to talk to someone on a call. If you can deploy this beyond chat as a channel, that, I think, is the holy grail.”
Conversica Chat will offer that flexibility, Kaskade said, by allowing ChatGPT interactions to happen across all those channels and even in-store digital signage. This will put brands in a better position to drive conversations that don’t simply answer questions but encourage purchases, contract renewals, expansion into new products and services and, ultimately, retention.
While revenue could become one of the most critical CX metrics for evaluating the return on investment of tools like Conversica Chat, Kaskade suggested there could be a new one: the volume “conversation qualified” customer interactions.
“Most CMOs and heads of marketing know that their lead scoring sucks, and that it doesn’t really work well,” he said, referring to approaches like determining marketing qualified leads (MQLs). “When you add the concept of conversation qualified, you know you’re referring to an exchange that went back and forth and that the customer was totally engaged. This could be a way of upping each of those existing key metrics across the funnel, where conversation qualified means the results they’re seeing are of higher quality.”
