“When you talk about AI, contact centers and customer experience, I think that’s the best use case to illustrate return on investment,” said Luiz Domingos. Every few seconds an agent saves in documenting the results of the call and every minute saved in gathering data about the customer and bringing it to the agent represents value.
Domingos is CTO at Mitel, a half century-old Canadian communications business. He had previously overseen the Unify communications software and services sweet at Atos and came to Mitel when it acquired Unify last year. We had earlier met with Mitel CMO Eric Hanson. He gave us an overview of the current business, which is invested in providing customized communications capabilities, across all channels, for complex businesses.
Not just a standalone suite, Mitel’s solutions also connect with other communications tools. For example, it recently announced a partnership with Zoom. “It’s very integrated,” said Domingos, “and we’re building applications and solutions that will create what we call a hybrid solution. In other words, we connect Zoom with the more traditional unified communications that Mitel has, because our customers invested a lot in those infrastructures and they’re satisfied with them. They’re not looking to replace them with some Zoom phone. We agreed to create a common client application that will be hosted under Zoom Workplace. Basically, telephony will be done by a Zoom client connected to our PBX system.”
The key term here is “unified communications,” the integration of a business’s customer (and employee) engagement channels. UC has been described as a business and marketing concept, but in simple terms it provides the technological background for customer experience.
When we asked about the role played by AI, he redirected us to Domingos.
Mitel’s AI journey
“We’ve been doing a lot on that front,” said Domingos. “At Mitel, we have been implementing what we call the ‘virtual agents.’” Essentially, these are combinations of AI assistants (helping customers) and AI copilots (helping reps). “They answer some calls, collect some data, provide the resolution of the call without going to a live person,” explained Domingos.
Mitel has been pursuing this kind of outcome for some time. “We had started that in 2019 with Google CCAI [Contact Center AI]. It was a little difficult to use, needed a lot of programming.” Generative AI, says Domingos, transformed everything. “It becomes way simpler to have an agent that can be a companion for the representative throughout the whole call and throughout the engagement. The agent assistant can resolve simple cases or accelerate the resolution.”
Mitel’s current virtual agent solution is called Talkative and it occupies the front end of the contact center. “So we have that in our products now. We have a solution called Talkative that simply does that, front end of the contact center call,” said Domingos. “It’s omnichannel, so not only voice but any other conversational channels like chat.” It provides answers for the chatbot or speechbot. Contact centers are loving it, said Domingos, because it means the human personnel can be more specialized.
Talkative for Healthspan
Healthspan is a U.K. vendor of longevity protocols, helping maintain health throughout the aging process. A big part of its business is supplying vitamins, minerals and health supplements. We spoke with Rob King, IT and facilities manager, based at Healthspan’s headquarters in Guernsey, one of the U.K. Channel Islands. Healthspan was already a Mitel customer, using MiVoice Business, a suite delivering voice, messaging, mobility, conferencing and other capabilities.
Healthspan was interested in the virtual agent to help meet its contact center challenges. “It is a massive part of our business,” King told us. “We have a 55-seat customer service center at Gosport on the south coast of England. It’s always been a really busy operation. We have a department that deals with sales orders coming in by phone; customer service is another big one — we really pride ourselves on our customer service so it’s really important that our customers have many channels to communicate with us.”
Previously it had a voice-only inbound contact center but that slowly evolved. “We have Talkative now, so we have a live chat facility on our website for customers to contact us,” said King. Human agents were already working in a Mitel environment to handle voice calls. “Talkative is integrated into that Mitel platform to offer live chat and AI functionality on our website,” King explained. “A speech bubble gives customers the option to live chat with us. Those chats go through into the back-end of the same Mitel system our agents are using to answer voice calls.”
The chatbot fulfills three functions. “It gives customers the option to live chat with agents in the contact center. It gives customers the option to look up order status based on an order number. The other option that we’ve just implemented in the last few months is AI functionality. This bit was quite exciting for us. It’s automatically looking at answers in a database we control and pulling back that information to give to customers. We sell over 300 products; a customer might ask what they should take for joint problems; it will automatically look at the repository of information we created and give answers to customers.”
ChatGPT is running in the background, but the answers are coming from Healthspan’s own knowledge base. The benefit? “It’s deflected these repetitive questions that are normally asked to agents to free them up to deal with live voice calls.”
And it has done so at an impressive rate, with Talkative resolving 90% of product queries after its first month of operation. Note, that’s not 90% of all calls, but specifically product queries, answered by a Talkative-powered chatbot Healthspan calls Product Professor. “That’s the name for the chatbot, something we came up with internally. We do make our customers aware at the end of the conversation that it’s an AI-generated response,” said King.
Dig deeper: Oracle aims to automate the complete customer service lifecycle
Meeting the knowledge base challenge
Using AI virtual agents to automate contact center responses could be disastrous if the AI is not trained on a reliable, comprehensive and up-to-date knowledge base. As we all know, if AI doesn’t have a good answer to a question, it’s capable of making one up.
Mitel faced that challenge internally. “We created an assistant application that learned from our products,” said Domingos, “and now whenever our service teams or channel partners have issues, they don’t call a hotline or write a ticket; they go to that chatbot and many times get the answers they need. We keep evolving it; we’re now nine months into the release.”
Mitel has sought to translate that experience into what its customers need. “When we create a contact center the idea is that we deploy the [AI-powered] chatbot and then offer a training option to the customers. Give us your data and we will train it. If our customer feels comfortable, we offer them a toolkit where they can train it themselves.”
As a vendor of health-related products and services, Healthspan takes responsibility for its own knowledge base. “Everything within the knowledge base has been written by a regulatory team employed directly by Healthspan,” said King. “Because we’re selling supplements and vitamins, there are regulations in terms of what we can and can’t say. That’s why we have an internal team that is constantly checking that the information is accurate and current before it goes live. The database is entirely controlled by Healthspan, so the AI response is only looking at certain URLs on our own website; it’s not reaching out to the whole of the web. I’m in charge of uploading the correct URLs into the Talkative platform. It’s quite a team effort.”
It took eight weeks before Talkative went line. “It’s taken off really well,” said King. “It was straightforward to implement and a lot of people are using it. “
Employee and customer experience
An improved employee experience is one desirable outcome of implementing AI assistants in the CX space. For Domingos, there’s ROI in reducing the size of your human rep base using the AI as an agent for handling the most simple cases. The human reps can devote more time to resolving complex issues. But this is not necessarily a case of AI stealing jobs. Working in a contact center can be stressful, there’s high churn, and, as a result, some say contact centers are understaffed in the first place.
As for the customer experience, it seems clear from Healthspan’s data that their customers are relying on Product Professor heavily. Of Talkative’s use more generally, Domingos says: “It promotes the contact center to a higher profile in supporting and aiding customers. For the customer experience: My cases get resolved faster, I don’t have to wait in line for 30 minutes, I don’t have to have a follow-up call.”
Dig deeper: Customer experience management in the age of agentic AI