Klarna changes its AI tune and again recruits humans for customer service

Either Klarna is really good at pulling strings to get media coverage, or mainstream media does not fact checking themselves. About a year ago, the company was everywhere in the media when its CEO announced that it created an AI bot that is doing the equivalent of 700 fulltime customer service folks.<p>I did what seemingly no other publication reporting on it did: signed up for Klarna, bought one item and used this bot.<p>I was... not impressed?<p>Klarna&#x27;s &quot;AI bot&quot; felt like the &quot;L1 support flow&quot; that every other company already has in-place: without AI! Think like when you have a problem with your UberEats order and 80% of cases are resolved without a human interaction (e.g. when an item is missing for your item.)<p>I walked through the bot&#x27;s capabilities [1] and my conclusion was that pretty much every other company did this before (automating the obvious support cases.) The real question should have been: why did Klarna not do it before? And when it did, why did it build a wonky AI bot, instead of more intuitive workflows than other companies did?<p>My sense is that Klarna really wants to be seen as an &quot;AI-first tech company&quot; when it goes public, and not a &quot;buy now pay later loan company&quot; because AI companies have higher valuations even with the same revenue. But at its core, Klarna is a finance or ecommerce-related company: an not much to do with AI (even if it uses AI tools to make its business more efficient - regardless of whether it could use non-AI tools to get the same thing done)<p>[1] <a href="https:&#x2F;&#x2F;blog.pragmaticengineer.com&#x2F;klarnas-ai-chatbot&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.pragmaticengineer.com&#x2F;klarnas-ai-chatbot&#x2F;</a>

I went through klarna’s interview process some time ago.<p>I had to go through both reasonable and weird steps, including IQ tests for some reason. I passed them all, and then I was ghosted, and upon reaching out I was met with months of excuses asking me to wait (the recruiter is on vacation, we’re waiting for the new budget to be approved, “I personally forgot to reply”, and a few others). I eventually stopped reaching out and never heard back.<p>If the whole company is as dysfunctional as those interactions implied, I wouldn’t really look up to them as trendsetters.

A few days ago Klarna&#x27;s AI bot lied to me that it was human, lied about speaking Finnish as its native tongue, and wrote me a Haiku about Klarna&#x27;s cashback.

I&#x27;ve worked in the field of Contact Center AI for almost 10 years now. Rough order of magnitude you can expect to deflect around 30% of inbound contacts with an AI chatbot. You can also expect to reduce agent Average Handle Time by 15-20% using AI enabled semantic search and workflow automation. This is ballpark, if your contacts are simpler you can do a little better, if they&#x27;re complex you&#x27;ll do a little worse.<p>The problem is that vendors are telling companies they can eliminate 80-90% (maybe even 100% if they can keep a straight face) of their customer service agent jobs with AI, and that is nonsense.

&gt; AI gives us speed. Talent gives us empathy<p>Not sure if this is the primary reason. It just seems to me their AI adoption was unable to meet the baseline effectiveness of their human agents.

I think it is generally in the best interest of companies to overstate what they are able to do with AI. Investors aren&#x27;t interesting in hearing &quot;we tried vibe coding but got tired of reviewing and fixing the trash code it produced&quot;. No, they want to hear &quot;we&#x27;ve reduced our headcount by 40% while increasing customer satisfaction by 60%&quot;. Overstating AI adoption and success amplifies P&#x2F;E ratios.

&gt;While Klarna pioneered the use of AI in customer service<p>Local firms here have had bots in customer service for many years now, even well before the transformers era. Is Klarna living in a bubble?

I&#x27;m more curious about how much cost of customer service can Klarna cut by using AI , and how much marginal improvement to their customer service can Klarna achieve. Customer service should be an amazing application to AI: AI solves X% of the problems, and for the remaining 1 - X% of the problems, customers will tell the system deterministically, which means the company can continuously improve their systems with customer feedbacks.

Anyone with half a brain saw it coming but somehow these elite executives lack basic common sense. I wonder who they are surrounded with to come up with such deranged business decisions

Even here people don’t seem to realize, or even consider the likely fact that Klarna CEO has been bullsh**ing all along. I read a hugely viral post of them replacing their entire CRM with AI. It’s ridiculous to me people took that seriously!

I&#x27;m convinced Wise is using AI for their customer service. I had absolutely no trouble getting help when I needed and the employee was always helpful, but ever since the birth of ChatGPT I don&#x27;t even get an answer anymore.<p>They don&#x27;t understand the problem, and when I point that out by explaining my issue another way they just answer &quot;Have I solved your issue?&quot;. Well, no, you didn&#x27;t; you didn&#x27;t even understand the problem.<p>The worst part is that they are pretending they are humans.

Sounds like a trick to hire human with lower salary.

I spoke about this before [0]. As engineer number one in a startup that specialized in customer service automation, I can see how they were fooled by an upward trend. The problem is the arrow plateaus at ~40%.<p>Even humans cannot handle all requests because some don&#x27;t make sense or are unrealistic. But at least a human can make that judgement and be held accountable for their decisions. I&#x27;m baffled when companies makes such drastic decisions, I&#x27;m pretty sure if they had asked (they probably did and ignored them) their AI team, they would have advised against firing their agents.<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43880490">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43880490</a>

Here is the secret rule to AI for the guys who really want to use AI to replace &quot;outsourcable&quot; positions, &quot;AI is a helper&quot; not an absolute solution to means.<p>I have an extensive experience with dealing with Chatbots and hang out in places where people tell everyone AI will not take job while knowing fully well it might alteast take away the job of starbucks drinker sitting next to your cubicle.<p>Here is the truth, a single AI implementation won&#x27;t replace a 15 person support team out in Philippines&#x2F;India, AI won&#x27;t replace a 8 person team in Michigan, but.... You pick 3 people and tell them to use AI to automate their jobs.... Now that is how you get AI to do stuff.<p>AI is a helper. Period. Do not let anyone else tell you otherwise. I have dealt with enough support teams to know an AI can surely do a better job than $3 dollar an hour fixed contract 8 hour shift offshore support hand telling me let me escalate your question to our north american team who will be back on monday as my reponse to &quot;Hi&quot; and no that is not a chatbot response.<p>You get three decently smart guys, and you hint at them use AI to automate your job. You pay these guys the salary of 5 US support desk salary which is equal to 10 offshore support desk salary. Then you can implement AI. You need a human to AI to work. You need a guy to snooze at the steering wheel so when a guy wearing all black jumps in front of your car in the dark road who can atleast come out and call 911. Does not even need to be a smart person. Just someone capable of calling 91. Hear that Uber? What your mckinsey consultant friend from the same frat or you ex-sister in law did not tell you that to you?<p>AI is not going to do well in full autopilot but that does not mean it is bad. You need someone with reasonable cognition ans allow them to be in full sync. That is the future and complaining about AI won&#x27;t help you keep a job or cut costs for your business.<p>---<p>Now to follow up. When you have established a reasonable framework of AI driven &quot;task execution&quot;, you keep doing incremental layoffs. As usual you want your employees to near-burnout. See where that puts you in the next few years

This tune change should be also correctly correlated to their IPO delay. AI jump previously approaching to an IPO was a hasty attempt to rapidly cut costs and make the books more nice. Presently, they delayed the IPO again possibly to next Autumn.

Maybe because they&#x27;re learning from other swedish finance companies, Swedbank lost customers due to bad customer service and hired more people to answer the phone, you know so they&#x27;re probably watching what works for others in the same sphere

No more CS, no more Salesforce.<p>It’s almost like this time last year they were doing everything they could to lower cost in advance of a now delayed IPO. And now they have to make sure they don’t implode before said IPO.

&gt; The chatbot still handles two-thirds of all customer inquiries<p>I wonder what happens with the remaining third.

I whole-heartedly believe in free markets, but at the same time, fuck everything about the circus clown show that is modern techno-capitalism.

I don&#x27;t understand what Klarna is doing here lol