• News
  • People
  • Long Read
  • Opinion
  • Weekend Wrap
SUBSCRIBE

Feature

Ultimate.ai pursues the perfect marriage of man and machine

The Helsinki-based startup’s sophisticated algorithm is capable of utilising big data to learn the rules of virtually any language.Ultimate Ai

Finland’s Ultimate.ai is in the business of creating superhuman customer service employees.

Reetu Kainulainen is uneasy with the idea that artificial intelligence and machine learning should be utilised to make customer service employees redundant.

“There has been a lot of talk about automation and productivity. It’s chilling talk. Customer service, after all, is a very unique organisation that has one-on-one conversations with customers even if we’re talking about a giant telecommunications company with tens of millions of customers,” he reminds.

“We don’t believe that artificial intelligence should replace customer service employees, but that it should be used to give them superpowers.”

Kainulainen is talking about Ultimate.ai, a Helsinki-headquartered startup that utilises artificial intelligence and deep learning to help companies to scale their customer service functions. He and his fellow co-owners decided to pursue the business idea after putting to use their unique mix of expertise in machine learning, text understanding and online customer support in a hackathon organised by Messukeskus.

“We won the hackathon and started listening to their challenges and observing how their customer service works. It turned out it was a one-plus-one-equals-three situation,” he tells.

Ultimate.ai is essentially utilising natural-language understanding in the customer service context. Image: Ultimate Ai

Focus on quality

The challenge presented to the participants was the same that continues to flummox companies around the globe today: How to make sure customer service employees are able to give each and every customer the attention they deserve?

Ultimate.ai has set out to tackle the problem by utilising a neural network to dissect historical customer service data and determine the nature, frequency and resolution time of customer queries.

“We are assisting customer service employees, who may have eight chat windows open simultaneously – a fact that not only causes stress but also erodes the quality of customer service,” tells Kainulainen.

Some queries are so simple and frequent that its product is capable of automating the responses entirely, such as queries about opening hours during public holidays. Some of Ultimate.ai’s clients have achieved automation rates of up to 80 per cent, but a rate that high requires a limited scope where the AI is used, according to Kainulainen.

“Our clients can decide which cases they want to automate, but just because you can automate something doesn’t necessarily mean you should,” he reminds. “We tell them that 20–40 per cent is a good result: it guarantees that only the simple cases are automated and the quality stays high.”

…and analytics

Other queries, meanwhile, are forwarded to customer service employees for the still inimitable human touch.

“We’re trying to understand the conversations and recommend responses to customer service employees in real time. The employees stay in control and can choose and edit the suggested responses freely before sending,” he tells, adding that the employees are consequently able to complete 20 per cent more sessions at a higher quality.

The process is reciprocal, as every decision made by a customer service employee – be it to select or edit one of the suggested responses – will help to improve the offering of Ultimate.ai.

The team is all about empowering customer service. Image: Ultimate.Ai

Kainulainen points out that automating customer service provides companies not only with productivity improvements, but also with a more in-depth understanding of the concerns and issues of their customers.

While employees, he explains, may struggle to provide a reliable estimate of the number of different types of customer queries, a chatbot can provide data on the current trends on a minute-by-minute basis, enabling the client to make adjustments, for example, to the user interface of its newly launched website

“Analytics is what we’ll be focusing on strongly in future,” he outlines.

Finnish comes in handy

The wide-ranging benefits have already been experienced by the likes of Elisa, Finnair, MTV, Posti and Telia. Kainulainen says the impressive references are testament of the product quality, rather than the sales prowess of Ultimate.ai, given that the three-year-old startup did not establish its first sales team until last December.

“It’s not because we have an ingenious sales strategy,” he chuckles, “but because our product simply fits the market.”

Ultimate.ai is relying largely on three selling points in its bid to take over the still fragmented customer service automation market: its compatibility with the most popular customer service management platforms, unique artificial intelligence that simultaneously supports and learns from employees, and powerful algorithm that is capable of understanding hundreds of languages.

“We have a running joke that our algorithms have to be very powerful because we started out in Finland, with the Finnish language. For once, the Finnish language is an asset globally. German is not a problem after mastering it,” says Kainulainen.

“A customer just took delivery of the product in Arabic, another in Chinese.”

By: Aleksi Teivainen
31.10.2019