Klarna saved $60 million and broke its company. The missing layer is what I’m calling intent engineering + 2 prompts to find yours

Source: Nate’s Newsletter Date: 2026-02-24 URL: https://natesnewsletter.substack.com/p/klarna-saved-60-million-and-broke

Summary

Nate uses Klarna’s AI case study — $60M saved, customer relationships damaged enough that leadership had to rehire humans — to argue for a third engineering discipline beyond prompting and context: “intent engineering.” The problem: AI optimizes for measurable metrics (speed, cost) rather than actual business objectives. The discipline progression is prompt engineering (what to do) → context engineering (what to know) → intent engineering (what to want). 95% of AI pilots fail to deliver impact while companies pour 21-50% of digital budgets into automation — a systemic alignment gap.

Implications

• Enterprise adoption thread. The Klarna pattern (AI succeeds at the wrong thing) is the most dangerous enterprise AI failure mode — harder to catch than outright failure because the metrics look good until the business consequence surfaces. Intent engineering — making organizational goals machine-readable — is the missing layer in most enterprise AI programs.
• AI economics thread. 95% of pilots failing while 21-50% of digital budgets go to automation suggests most enterprise AI spend is being wasted on capability without alignment. The companies that crack intent engineering will have dramatically better ROI than the market average.
• Watch: Whether “intent engineering” becomes a formalized discipline, and which enterprise AI platforms build explicit support for capturing and encoding organizational goals rather than just task instructions.