Case Study · AI Adoption
Klarna's AI assistant handled 2.3 million customer conversations in its first month
What happened
Klarna deployed an AI Assistant built on OpenAI's technology. In its first month, according to figures published by Klarna and OpenAI, the assistant handled 2.3 million conversations - about two-thirds of customer service chats - doing work Klarna estimated was equivalent to 700 full-time agents. Average resolution time dropped from about 11 minutes to under 2, and repeat inquiries fell 25%. Klarna estimated the financial impact at around $40 million in 2024.
Klarna also gave all employees access to ChatGPT Enterprise, with the company reporting around 90% daily usage. It didn't treat AI as a single experiment - it rolled it out, watched how it got used, deployed it into a real workflow, and measured the results. Note: the $40 million and 700-agent figures are Klarna's own estimates, not an independent financial audit - worth framing as "Klarna estimated," not as independently confirmed.
The business problem
Turning AI experimentation into a measured, company-wide rollout rather than a scattered set of individual pilots.
Why it worked
- Roll out, observe usage patterns, then commit - in that order
- Attach real metrics to the rollout from day one
- Company-wide access plus a governed workflow beat either alone
- Treat self-reported vendor figures as estimates, not audited results