Harvey is the fastest growing legal tech company in history.

Winston Weinberg joins me on this week's Legal Innovation Spotlight and shared Harvey's origin story and lessons learned long the way.

He was vulnerable about things they got wrong and it allows his humanity and his humility to shine through.

It's a look more CEOs should embrace....including me.

Every CEO's journey involves learning by doing and that comes with making mistakes.

Embracing those mistakes is how you learn.

Owning them is how you earn respect.

Most people look at Harvey and think they were the golden child from day one.

But it started with a Reddit experiment.

Winston and his co-founder Gabe grabbed 100 landlord-tenant questions from r/legaladvice.

They built a chain-of-thought prompt using actual statutes.

Then they asked three attorney friends to review the outputs.

The instructions were simple.

Thumbs up if you'd send this to a client with zero edits.

Thumbs down if you'd change anything.

On 86 of 100 questions, all three attorneys said thumbs up.

No edits needed.

This was mid-2022.

GPT-3 had a terrible UI.

No instruction fine-tuning.

It would autocomplete your sentences if you forgot a period.

They cold-emailed Sam Altman and OpenAI's general counsel that week.

Subject line was something like "AI assistant lawyer."

OpenAI responded.

By July 4th they were pitching the C-suite.

The rest is history.

The lesson here isn't about luck.

It's about testing with real practitioners on real questions before building anything.

86% wasn't a projection.

It was proof.