Are Entire Companies Experiencing AI Psychosis?

The Warning

Mitchell Hashimoto (creator of Vagrant, co-founder of HashiCorp) posted to X about companies operating under what he calls 'AI psychosis' — and the thread hit 1,650 points on Hacker News. His concern: entire companies are now run by teams who believe AI agents can fix bugs so quickly that shipping broken software is fine.

What AI Psychosis Looks Like

The psychosis, as Hashimoto describes it, centers on an 'MTTR is all you need' mentality: mean-time-to-recovery via AI agents is so fast that system reliability no longer requires investment in resilient architecture. Bug reports go down. Test coverage goes up. Everything looks healthy by local metrics.

But Hashimoto argues that 'systems can appear healthy by local metrics while globally becoming incomprehensible. Bug reports can go down while latent risk explodes. Test coverage can rise while semantic understanding falls. Changes happen so fast that nobody notices the underlying architecture decaying.'

The Infrastructure Parallel

He draws on the MTBF-vs-MTTR debate that played out during the cloud automation era. At the time, teams argued about whether automation-enabled recovery speed made resilient system design unnecessary. The lesson from that era: MTTR is valuable, but you cannot automate your way into ignoring systemic fragility.

That same argument is now happening across the entire software development industry — and possibly society at large.

Why It's Hard to Argue Against

What Hashimoto finds most concerning is the impossibility of the conversation. Raising this concern with people in AI psychosis mode gets responses like 'we have full test coverage' or 'bug reports are going down' — local metrics that sound correct but miss the global picture entirely.