Mitchell Hashimoto Noticed Entire Companies Under 'AI Psychosis.' That Observation Hides a High-Value Service Business: Helping Organizations Diagnose and Recover from AI Decision Abdication.

Mitchell Hashimoto built HashiCorp.

He knows how fast-moving engineering organizations work.

So when he says "I strongly believe there are entire companies now under AI psychosis," that is not a casual take.

It is a pattern observation from someone who has watched a lot of teams ship things.

The Hacker News post hit 376 points and 142 comments within hours.

What AI psychosis actually looks like in a company

The behavior is not just "using AI a lot."

It is a specific failure mode: outsourcing judgment without maintaining oversight.

Signs that came up in the thread:

• Shipping code without understanding it because "the agent will fix bugs quickly"
• Building systems that no human can fully trace or reason about
• Accepting AI output as a decision rather than a draft
• "If you just prompt the AI and believe what it tells you then you have AI psychosis"

The distinction Mitchell drew is important.

He uses AI daily.

The difference is: he evaluates the output. He does not treat it as a verdict.

Why this is a service business, not just a warning

Every fast-moving startup with active AI tooling has some version of this problem.

They moved quickly.

They shipped.

They used agents.

Now they are not fully sure what is running their stack.

That is not a failure of values.

It is a predictable outcome of speed without process.

The people building these companies are smart. They just need a structured way to see their own blind spots.

That is the audit.

What to offer

A fixed-scope engagement: an AI Psychosis Audit.

You are not selling fear.

You are selling visibility.

The deliverable is a clear picture of where the company has blind trust in AI output — and a prioritized plan for what to verify, what to checkpoint, and what to rebuild with better human oversight.

The clients are easy to identify.

Any founder who reads the HN thread and thinks "we might have some of this" is already your buyer.

They are not in denial.

They are just unsure where to start.

The pitch in one sentence

"We give you a two-week audit that shows exactly where your company is accepting AI outputs as decisions — and a roadmap for getting oversight back."

That is it.

No jargon.

No evangelizing.

Just visibility and a path forward.

The follow-on product

The audit is the door.

The sprint is the business.

A 30-day AI Hygiene Sprint that installs the specific process changes, verification checkpoints, and team habits that turn AI psychosis into productive AI use.

You are not asking companies to use AI less.

You are helping them use it better.

Mitchell said it plainly: the problem is not using AI.

It is outsourcing judgment without knowing you did it.

Name that clearly, offer a structured way out, and you have a business.

Sources:

https://news.ycombinator.com/item?id=48153379