Jul 17, 2024 · 55m · big-technology
Did Klarna Really Automate 700 Jobs With AI? — With Sebastian Siemiatkowski
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Klarna CEO Sebastian Siemiatkowski joins the Big Technology Podcast to discuss how the fintech giant leveraged generative AI to automate customer service, overhaul enterprise workflows across marketing and legal, and restructure its workforce through natural attrition.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Alex holds 20.4% of the talking time here. How this is scored →
Alex as informed peer 4.9 Guest teaching 4.2 Guest disagreement 2.4 Alex pushing back 4.8
speaking balance: gold is Alex, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 25:44 Guest directly challenges host premise on marketing orchestration
When Kantrowitz asserts that core marketing functions cannot be handled by AI, Siemiatkowski immediately interrupts with 'Ah, you think so?' before explaining how breaking tasks into granular units enables AI orchestration.
Hardest push from Alex ▶ 19:02 Host presses on whether AI is papering over previous operational gapsKantrowitz directly challenges Siemiatkowski on whether Klarna's dramatic headline metrics were driven by actual AI power or merely catching up on basic phone automation they lacked.
Biggest teaching moment ▶ 42:16 Guest educates host on PR automation beyond sentiment analysisAfter Kantrowitz dismisses PR AI use as trivial sentiment reading, Siemiatkowski reveals the system actively monitors 40 daily articles to flag factual inaccuracies against centralized enterprise data.
Alex holds their own ▶ 12:07 Host conducts live UI test showing bot limitationsKantrowitz demonstrates first-hand investigation by testing the Klarna app live, reading back the bot's response to prove it only provided manual instructions rather than autonomously executing the action.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Alex as informed peer | Guest teaching | Guest disagreement | Alex pushing back | Why |
|---|---|---|---|---|---|---|
| Debunking Rumors and Developing the AI Customer Service Agent | 5 | 5 | 3 | 6 | Kantrowitz opens by skeptically pressing Siemiatkowski on Klarna's claim of replacing 700 reps, noting the coincidence of their prior 700-person layoff. Siemiatkowski clarifies that the layoff was two years prior and explains the organic development of their internal dispute co-pilot. | |
| Impact on Outsourced Customer Service and Vendor Reactions | 4 | 4 | 2 | 4 | Siemiatkowski details how Klarna's lack of legacy phone trees made AI impact larger, clarifying that the 700 reduced headcount came from outsourced BPO vendors. Kantrowitz presses on vendor reactions and contract reductions. | |
| Resolution Speed and In-Chat Task Automation Mechanisms | 6 | 4 | 2 | 5 | Kantrowitz shares his live test of the Klarna app, pointing out the AI provided manual navigation instructions rather than autonomously completing the refund. Siemiatkowski explains that in-chat task automation relies on specific UI widgets built into the app. | |
| Managing AI Hallucinations and Quality vs Human Agents | 5 | 4 | 2 | 4 | Kantrowitz asks about hallucinations and customer bot exploits, analogizing acceptable error rates to autonomous vehicles. Siemiatkowski explains that human agents also make errors and that Klarna benchmarks AI performance to human parity. | |
| Reasons Behind Klarna Public Transparency on AI Impact | 4 | 3 | 2 | 5 | Kantrowitz asks why Klarna would publicize its operational advantage rather than keeping it proprietary. Siemiatkowski admits a mix of proactive regulatory communication, societal awareness, and company self-promotion. | |
| Distinguishing Generative AI Impact from Baseline Process Automation | 6 | 4 | 3 | 7 | Kantrowitz presses directly on whether Klarna is exaggerating gen AI's power to mask a previously unautomated customer service setup. Siemiatkowski concedes that the savings are roughly 70 percent gen AI and 30 percent baseline automation catching up. | |
| Transforming Multilingual Marketing and Copywriting with AI | 5 | 4 | 2 | 5 | Kantrowitz questions Klarna's claim of 80 percent AI-written marketing copy and asks if unreviewed AI copy goes live. Siemiatkowski details localization across 20 countries and multi-agent writer-reviewer configurations. | |
| AI Employee Interviews and Deconstructing Complex Workflow Tasks | 4 | 6 | 4 | 4 | Kantrowitz doubts AI can handle strategic marketing orchestration, prompting Siemiatkowski to push back with an example of an AI-driven qualitative employee interviewing system that deconstructs complex qualitative workflows. | |
| Creative Boundaries of LLMs and Automated Image Workflows | 5 | 4 | 3 | 6 | Kantrowitz shows stock imagery in the live app to challenge claims of full visual automation. Siemiatkowski distinguishes between creative, non-average human copy and multistep automated image generation pipelines for catalog assets. | |
| Evolution of Financial Automation from Scripting to Proactive AI | 5 | 4 | 1 | 3 | Siemiatkowski recounts Klarna's history with banking screen-scraping and robotic process automation via Sofort, framing modern generative AI as the next step toward proactive financial optimization. | |
| Enterprise ChatGPT Adoption and Practical Utility Across Teams | 4 | 3 | 2 | 3 | Kantrowitz questions whether Klarna's 90 percent daily ChatGPT Enterprise usage is sustainable. Siemiatkowski contrasts generative AI utility with blockchain and emphasizes building custom internal assessment tools. | |
| Media Error Detection and Overcoming Enterprise Data Silos | 5 | 6 | 3 | 5 | Kantrowitz asks why PR sentiment analysis needs AI when humans can read it. Siemiatkowski explains that the AI actually automates factual error detection across 40 daily articles by indexing structured internal company data. | |
| Accelerating Legal Workflows and Contract Drafting | 6 | 4 | 2 | 5 | Kantrowitz asks about legal contract drafting and probes Klarna's 85 percent valuation collapse during the fintech reset. Siemiatkowski contrasts BNPL installment models against revolving credit card debt and outlines planned workforce reduction via natural attrition. |
Statements from this episode (25)
Siemiatkowski: Viral 700-person layoff figure was entirely coincidental to AI gains
“So the layoff of like the comparison to the seven under layoff is actually a misquote by a news magazine online. It's not accurate. It was two years ago when we had to you know, change the amount of investments we were doing, we had to make layoffs and it just…”
Klarna's AI dispute co-pilot cut a 30-day backlog to zero
“The disputes at us had always been, like, there was a backlog of, like, 30 days. It's a quite complex matter. It's always frustrating to us because consumers want to get answers very quickly, and we need to collect a lot of information, so it takes a little bi…”
Klarna's AI customer support agent absorbed the workload of 700 human reps
“Now the difference was when we Got this customer service AI agent to reach a level where it actually served a lot of errands, and on a satisfaction level equal to what human agents many times did, and we took it live, the number of errands that our human agent…”
Klarna cut outsourced customer service spend by $40 million using AI
“We on average would have like a about 3000, it depends on because you have to remember also like our, just like Amazon, We have much more transactions around Christmas because we're very online, you know, so there will always be variations in these numbers, bu…”
BPO vendor lost over $1B market cap after Klarna's AI tweet
“Well, they were not very happy because we tweeted about it, and a few of them had very severe implications on their market cap, because, like, one of them lost, like, over a billion dollars in market cap on the stock exchange.”
Klarna's AI cut customer support resolution time from 14 to two minutes
“When a human started a conversation with another human to resolve a single task took on average, 14 minutes, right? Just because of those delays that happened because people aren't really actively talking to each other all the time. Somebody said, Oh, let me g…”
Many Klarna customers immediately demand human agents when encountering AI bots
“There's also a huge amount of customers The first thing they write when we expose the AI agent to them is agent, right? That's the first thing because a lot of people have had so much bad experiences with these AI bots that they just want to talk to a human, r…”
Klarna considers AI chatbot error rates acceptable if on par with humans
“So what we simply do is we read a lot of these transcripts on a continuous basis, and we do continuous quality checks to ensure that the error rate is not higher for the AI chatbot than it is for our human agents, and if we see that they are at least on par, T…”
Publicizing AI results boosted Klarna's inbound startup partnerships and hiring
“What we've seen as a consequence of sharing this is that more of the AI startups wants to work with Klana because we're regarded as, you know, a thought leader or somebody doing something exciting in the space. We see more people want to work here, you know, e…”
Siemiatkowski estimates 50% to 70% of Klarna's efficiency gains come from AI
“My belief though, if you ask me, is that like, I don't know, like maybe 70% is AI and 30% is automation or fifty-fifty. Yeah, but I still think it is that much, actually. That's still my belief, right?”
Siemiatkowski: Within two years, AI customer service will outperform legacy automation
“So I think that, like, even though maybe the first iteration would only have been a substantial improvement compared to some other companies, if you give it one or two more years, it will definitely be an improvement versus what any other company is doing with…”
Klarna built an internal AI tool to conduct conversational employee engagement surveys
“One of the first AI applications we actually built internally, Was, and this was again, just like an idea that we just did. It, it's not Klarna's core business, but one of the, it's just a concept that we wanted to test... So what we decided to do is we create…”
Siemiatkowski: LLMs regress toward the average and fail at breakthrough marketing copy
“If I want some amazing copy that, you know, articulates what Klana does that's different, I would not rely on ChatGP to do that. I would rely on a human because it, a human has a bigger, is much better at kind of being far-fetched and do something crazy and ou…”
Klarna uses AI for in-app category images, but not major brand campaigns
“Would I do that for my, you know, Super Bowl campaign? No. Would I do that for category pictures in the app that are just there to say, hey, we have shoes, we have this? Yes. That we're already doing and applying.”
Klarna processes $100 billion in volume, $30 billion via bank scripting
“We process, we do about a hundred billion dollars worth of volume. We do thirty billion volume on debit through that solution, through basically scripting.”
Siemiatkowski: High consumer switching costs drive legacy banking's excess profits
“A lot of the banking profits are generated due to the fact that the switching costs are so high and we're not willing to switch. And so the competitive pressure is actually lower than it's perceived to be.”
Siemiatkowski predicts 10-year-olds won't need driver's licenses due to autonomous vehicles
“I personally believe that self-driving cars at some point in time will happen, but I don't know when my bet is my daughter is now 10, and I have always said, I don't think she's going to get a driver's license, but that's eight years out, right?”
Siemiatkowski: Bitcoin and blockchain failed to solve real consumer problems
“When, like, when Bitcoin came along and these technologies, like, I tried that as well. I personally didn't find the technology to solve a real problem. In my opinion, I didn't see how it was gonna help my mom prefer using Klarna Over something else. Like, I m…”
Klarna analysis finds about 20% of its press coverage contains factual errors
“I think we've analyzed it about. 20% of the articles that are written about us contain factual errors. They're incorrect.”
Klarna uses AI to detect media errors and draft journalist corrections
“Having it assess those articles and identify those errors, and then even sometimes draft an email to the journalist to ask them to correct it. It's like a nice, it's a nice thing to avoid that manual work and spend time on something.”
Klarna was consistently profitable from 2005 to 2018 before expanding to the US
“Klona was profitable from 2005 to 2018. So we had a history of kind of running this a little bit differently than most tech companies just like burn money. We have been profitable, but then when we came to the U.S., we invested heavily and that meant that we w…”
Siemiatkowski: US banks earn $5.50 per $100 spent versus Klarna's $2.50
“So if you look at the total cost of payments in the U S and you look at both what the bank is earning, By interest on your revolving plus the merchant fees. It's actually a crazy. 5.5 dollars on a hundred dollar spend. And we earn less. We do about two and a h…”
Klarna's average balance is $150 versus $5,000 on US credit cards
“The average outstanding credit card balance is 5000 dollars. The average outstanding balance on Klarna is a 150.”
Klarna shrank headcount from 4,500 to 3,700 via attrition and a hiring freeze
“We haven't been hiring since September. And since we, as many tech companies have a natural attrition rate, we're about, you know, 20% leave on an annual basis. So people stay about five years, which is kind of typical for tech companies. This means that we ar…”
Klarna's total employee cost will fall while compensation per employee rises
“Our total employee cost will go down, but our cost per employee will go up. Right? So that is basically a commitment to our employees that like, They will benefit from this in seeing higher salaries and more equity shared with them, which is what we've done. A…”