AI & Customer Support: Klarna replacing AI with real people

In this latest post we explore the use of AI in customer service with Aidan O’Shea the CEO and Co-Founder of Otonomee. We examine the recent comments from Sebastian Siemiatkowski, the CEO of Klarna in relation to this topic.

The context for this interview was rooted in comments from Sebastian Siemiatkowski, the CEO of Klarna reported on Bloomberg on the 8th of May last year, Klarna Slows AI-Driven Jobs Cuts With Call for Real People.

In the article, he is quoted as saying that he now believes the company’s AI-driven cost-cutting plan at Klarna has “gone too far” and he “wants to recruit a new division of remote workers in rural and student populations to bolster customer service operations”.

The article reports that Siemiatkowski now believes it’s “critical” that a human always be available for customer service.

‍“As cost unfortunately seems to have been a predominant evaluation factor when organising this, what you end up having is lower quality,” he said. “Really investing in the quality of human support is the way of the future for us.” (source Sifted)

In some ways, his comments align well with the view of Aidan O’Shea and Otonomee who has consistently argued that great CX needs a combination of both human and AI agents. Furthermore, given that Otonomee is a remote-first company, the reference to ‘remote workers’ was also of interest, a topic we explore further below.

There have been some interesting comments in the press this week from Klarna CEO, Sebastian Siemiatkowski, in relation to support and AI, Aidan. Can I start by getting your views  on the comments?

I thought the comments were very interesting. When Klarna came out initially talking about their success deploying AI and how it was displacing 600 or 700 jobs, it was big news. 

Klarna partnered with ChatGPT maker OpenAI in 2023 and launched an AI customer service assistant using its technology a year later. The company said last year that AI was doing the work of 700 customer service agents. Source CNBC

Many people took it as another signal of the death of customer service managed by employees (or humans), and that “the machines were taking over.” The Klarna announcement at the time was one of the clearest signs yet of the transformative potential of AI in a customer support context. It is very interesting that it didn't take him long to come back and say “we've actually gone too far.”

‍Siemiatkowski said in a separate Bloomberg interview last week that the company will soon look to recruit more human customer service agents to work ‘in an Uber type of setup.’ He acknowledged a full tilt toward AI-based support roles resulted in lower quality work. Source CNBC.

As ever, the balance in terms of AI and employees working in support is somewhere in the middle. At Otonomee, we’ve always felt that the best operation is made up of the best combination of people and technology. It isn't one or the other. It's a combination of both that is going to work. So to be honest, I wasn't surprised to hear these statements. In some ways, I was glad to hear them because there's a lot of hype around AI. It has amazing potential, and we're seeing it gain more traction, especially in relation to automation. But actually, the benefit is automation. The benefit is making existing support agents better, whether that is more responsive or efficient, or in giving them the ability to deliver a better service by sourcing information faster. But it’s still the human who has the ability to be empathic, who can listen and communicate with the customer on the other end.

Or as a Klarna spokesperson declared:

‍"AI brings speed. Human talent brings empathy. Together, they let us deliver service that’s fast when it matters, and personal when it counts," they said. "It’s not AI versus humans – it’s AI and humans, working together. That’s where the magic happens." Source Sifted

At the heart of the debate is the narrative related to AI as a complement versus AI as a substitute. What we’re seeing from his quote is a great example of AI as a complement. Take his reference to the word empathy. There's this sense that AI is not good at empathy, and that we should let AI do what it does well, and layer on the human bit around it. Can you elaborate on that concept?

Customers don't want to reach out to customer support or service; they'd rather not. None of us wants to be calling or emailing or chatting about a problem we have. We just want things to work. If I'm going to the trouble of writing an email, submitting a question into a chat, or making a call, something is annoying me, or something probably has gone wrong for me. With some issues, it is not as simple as, “I can ask a question and I get an answer back.” There is often something else going on, and this is impacting me in some other way. In these scenarios, people want to speak to someone about “their issue.” They want to have human interaction. And of course, that can be a great opportunity for businesses to turn that initial negativity into something really positive. That is delivering a great customer experience. And this is a key point. Sometimes people getting in touch just want an answer. But other times they're looking for that human connection, someone to understand, someone to listen to them, someone to tease out “their issue.” In these circumstances, they’ll want the ability to speak to someone directly.

There was another part of the quote that I thought was very interesting, and I'd like your take on it, because it was very relevant to Otonomee? Siemiatkowski was quoted as saying he wants to recruit a new division of remote workers in rural and student populations to bolster customer service.

Why do you think he's mentioned the word rural there, and why is it important?

I can only give him my opinion. But it is probably because he wants access to good people. He says he wants empathy and also questions the quality of the service and the quality of the answers. So he wants to elevate that quality of the Klarna service offering. He’s essentially saying, I want to find better people and students. This implies that they need to be smart, they need to be educated. I think he's saying rural, because he can hire from a wider talent pool. He's not hiring in cities that are overpopulated, expensive, and where talent has a choice. Most cities in Europe will likely have several big brands, be that a Google or an Apple, that compete for this talent (naturally winning it in many instances). Whereas in rural locations, you don’t have that. So he's clearly identified that he can get good talent at a reasonable cost, and he wants to raise his bar in terms of the quality of their support offering. So I think that's why he's saying it.

Isn't that very consistent with the philosophy of Otonomee too, right? Remote working is very important to Otonomee.

Yes, indeed. Most of the senior leadership team have worked in BPOs for 20 years plus. We've been in cities, big cities, big buildings, all the time. When we set Otonomee up it was to solve the problem inherent in having a physical presence in a big city. There was a real war for talent. It was becoming clear that it was getting much harder to attract and retain good people. We realised we can actually get great talent and access to a much wider talent pool if we were remote-first. Not only did it help with access to better quality candidates but it also helps with retention. 

How are you deploying AI at Otonomee? 

We deploy it across most of our clients. Efficiency is key. There is an opportunity for automation in AI, in some contexts where you don't need people to be involved. So, for example, if you've a question that has been asked before, there's no reason why AI can't understand your question, find the answer, and get it back to you. Of course, the key point here is that there are other instances when AI alone won’t be sufficient.

The big opportunity for AI is to help solve standard issues quickly. When the issue is a little more complex or urgency or sensitivity comes into play, and where maybe empathy is needed, people are going to want to talk to a ‘real human’. 

In these scenarios, AI still has a role to play, as the customer support agent needs to search through data. They may need to search the previous five emails you might have sent about your query to really understand your case, because it may be complex. There's something happening here with your plan, your coverage, you know, your usage - whatever the case may be and you need the wider context to address the issue. So the support agent has to search through previous notes and records. This brings the following intro play - sourcing the notes, reading them, understanding them. That takes a lot of time. Whereas the AI can help in a scenario like this, not alone in quickly finding relevant notes, but also in terms of summarising it and presenting it back in a way that can help the agent to resolve your query in a timely manner. 

So in this scenario it is making the support agent smarter and more efficient. It is helping them to understand the background of your issue, the context of your issue, which then helps them to be more empathetic, more understanding. So even in a human-to-human context, it can help agents address issues more efficiently and help them to be faster to get to a resolution. 

It strikes me that a specialist, outsourcing provider, like Otonomee is well placed to take advantage of AI because it spans multiple clients, and you can cross-pollinate with best practices. 

Everybody's learning here, because it's all new. So Karna is learning, and has learned a lot from their in-house experience of deploying AI.  

We have the benefit of learning across a lot of different clients that are trying to utilise AI to support their clients more efficiently. We're probably learning a bit faster, but it is also because we are making more mistakes (or course corrections). We have the benefit of seeing what works across multiple clients, and we then have the opportunity to share the learnings with others as well. 

One final point on Klarna. When I first heard that they had managed to replace 600-700 roles with AI, my immediate thought was less about the AI and more about operational efficiency. 

Did they really need that many roles in the first place? 

Sometimes the real issues are about more effective operations, better use of data, or enabling your colleagues to be more effective so you don’t need your support operation to grow to that size in the first place.

About Aidan O'Shea

Aidan has over 20 years experience in outsourcing having supported many of the most iconic brands in the world. Aidan co-founded Otonomee in 2020 to address the challenges and weaknesses of the traditional BPO model. As Otonomee's CEO, he has scaled the company to over 350 remote employees in Europe, North America & Asia, capitalizing on the opportunity created during the pandemic around remote working, the ability to hire from a wider talent pool and to offer employees a greater employee experience.  

Former Managing Director of Voxpro (2014-2020), Aidan led the growth of the company to 6,000 employees globally along with the acquisition by Telus Digital, a Canadian Telco.

About Otonomee

Otonomee is an award-winning Customer Management Outsourcing business that delivers scalable, tech-driven, people-focused solutions. Leveraging our remote-first operating model, we optimize customer interactions, reduce costs, and increase efficiency. Our focus on remote work, automation, and data-driven insights makes us a next-generation partner for leading brands and scaling companies.

‍