Amazon is determined to use AI for everything – even when it slows down work
TL;DR
Amazon is pushing corporate employees to use internal AI tools like 'Kiro' – even though they frequently hallucinate and generate flawed code.
Key Points
- Developer Dina from New York now spends more time fixing AI mistakes than writing code herself, sometimes reverting all changes and starting over.
- Employees report increased surveillance pressure and more work overall, not less.
- The pattern: AI is deployed to demonstrate efficiency gains, while the quality-control burden quietly shifts onto staff.
Nauti's Take
'We use AI for everything' sounds like progress, but it is often just performance for investors and boards. The Kiro example is a classic deploy-before-ready problem: the pressure to demonstrate AI adoption trumps the question of whether the tool actually helps.
What gets ignored: bad AI creates new overhead – debugging, distrust, demotivation. Amazon should ask itself whether 'AI-first' is genuinely a competitive advantage or just an expensive way to produce the same output with more friction.
Context
Amazon is not just any company – it is one of the world's largest employers and a benchmark for tech industry trends. If even internally developed AI tools are reducing rather than increasing productivity, it challenges the core assumption behind many AI investments. What makes this especially significant is that the negative effects are not being communicated openly but are leaking out through employee reports, pointing to a gap between management narrative and lived reality.
Sources
