If you're locked out of your account at 2am, does a chatbot really cut it?
Thinking of automating customer support?
Klarna did. Now they’re undoing it.
➤ Imagine you’re Sebastian Siemiatkowski, Klarna’s CEO. You see a 85% valuation drop in 12 months, investor pressure, and a market downturn… You really need to cut some costs.
➤ Customer service is always the first to fall. All those smiling humans with headsets? Too expensive. Too slow. Too unpredictable.
➤ We’ve seen it before, in boardrooms across industries. It’s treated like a cost centre, not a differentiator.
➤ The promise of replacing the team with AI: cheaper, faster, and infinitely scalable. Even your CFO starts smiling.
➤ AI-first headlines? Even better. You look bold, efficient, and tech-forward - the kind of story investors love to read.
Sebastian Siemiatkowski – “This AI breakthrough in customer interaction means superior experiences for our customers at better prices, more interesting challenges for our employees, and better returns for our investors.”
➤ On paper, it created a dream dashboard. In the first month alone: 2.3 million chats handled, average session time slashed from 11 to 2 minutes.
➤ But “efficient” doesn’t mean “effective.” You’re not optimising a pipeline - you’re helping people with problems.
➤ No chatbot, no matter how large its dataset, can console a panicked customer locked out of their account at 2AM.
➤ That’s why “customer” comes before “support.”
➤ No surprise reviews were brutal. I mean zero-star brutal:
Krysteena, in G2 reviews – “They have the WORST CUSTOMER SERVICE (cut and paste answers with redirect to their phone line)”
➤ Reversing a bad call is hard. Admitting it was a bad call? That takes some ego surgery.
➤ They tried - for 3 years - to improve it, until they finally just hit undo and started hiring people again.
Sebastian – “As cost unfortunately seems to have been a too predominant evaluation factor when organising this, what you end up having is lower quality;”
➤ No kidding.
Sebastian – “Really investing in the quality of the human support is the way of the future for us.”
➤ LLMs don’t understand human emotions the way humans want - yet. Today, that makes it a bad place to cut costs.
➤ With so much potential for AI replacement, here’s the question leaders must confront: are you removing friction? Or just externalising it to the customer?