A lot of people want to build vertical AI startups, but most of us come from engineering/product backgrounds… not healthcare, logistics, finance, hospitality, etc.
So the question becomes:

How do you validate a workflow-heavy domain you’ve never worked in?

Some YC founders even take jobs inside these industries just to feel the pain firsthand; honestly, totally worth it. Recently, I followed a friend to the hospital and found myself in full detective mode: watching how nurses hit their keyboards, how doctors shuffled from room to room. I thought to myself that there might be a billion-dollar idea sitting right there on the hospital floor, and I wanted to go back home with that and not a cough.

At one point, I noticed a doctor jumped on a video call with a human interpreter just to translate a Spanish-speaking patient’s explanation of symptoms, and I was like, isn’t someone supposed to be building this already?

I started thinking: could this same detective approach help me find real workflows to optimize? That’s exactly how we found a two-week-long workflow at a major restaurant brand that Layout now completes in under 2 minutes, all by talking to the former CMO of Chipotle.

Here’s what I do:

After that hospital moment, the next question for me was always the same:
“Okay… but is this a real problem or just something I happened to notice?”

And that’s where the real work begins.

Whenever I get curious about a problem, I start by asking myself: Who would know the truth better than anyone else?

Every vertical AI idea has a “living encyclopedia”,  someone who has spent years inside the workflow you’re trying to understand. Sometimes it’s a CMO who’s lived through the chaos of marketing ops, sometimes it’s a Head of Nursing, and sometimes it’s an engineer who built internal tooling at Fortune 500 companies.

For Layout, that person was the former CMO of Chipotle. He knew the daily grind of coordinating local marketing across thousands of stores. Talking to him helped us validate the workflow and identify exactly what we could automate.

Cold messages work, but sometimes they can be slow.
What I’ve discovered instead is that many of these executives already want to talk,  just not in their inbox.

Platforms like intro.co, LinkedIn Services, and topmate.io turned out to be my cheat codes. The people on these platforms are already in “helping mode,” and many of them donate their earnings to charity, which removes the weird sales dynamic.

Because of these platforms, I’ve ended up on calls with CMOs and several Directors inside several Fortune 500 companies, often within a day or two.

Suddenly, you’re not “some random founder in their DMs.” You’re someone they agreed to talk to. That alone changes the quality of insight dramatically.

These calls are short, usually 15 to 30 minutes. If you want depth, you can’t show up asking basic questions.

So before the meeting, I always send a list of things I’m trying to understand:

I didn’t expect this to make a huge difference… but with Layout, asking the right questions was how the marketing executive at Chipotle walked me through a workflow that took their team two weeks to complete manually, and I realized Layout could automate the same workflow in just two minutes.

If someone is local, I always push for an in-person meeting. LinkedIn Services makes this easy since the default mode for the conversation isn’t an online call.

The best conversation I’ve ever had about an idea was a two-hour coffee with someone who’d built search systems across several public tech companies. He walked me through the entire history of the industry, the failures, internal politics, constraints, workarounds, everything.

There’s something about sitting across from someone that unlocks the stories they’d never say on a Zoom call (coffee + staring each other in the eye is magical).

Of course, this whole approach has its downsides.

First, an executive might not be the person living the problem you’re solving; some lower-level staff might be the person doing the grunt work, so this might actually not help you at all.

Second, it can get expensive. Some of these execs charge $300–$500 for 15 minutes, and a few go even higher.

Also, some people are too nice. If you pay someone that much, they sometimes feel obligated to validate your idea. They want to be encouraging. They want you to win. You leave the call feeling great… but with zero new information.

Then there’s the exact opposite. You meet people who have spent years fighting internal fires, and the trauma has turned them into professional pessimists.

They will tell you why your idea is impossible, why it won’t scale, why AI can’t handle edge cases, why this exact thing failed in 2017, and why you should go do something else.

And honestly?
Sometimes they’re right, but sometimes they’re just scared.

This is where Patrick Collison’s line has helped me a lot:

“Pessimists sound smart. Optimists make money.”

Your job isn’t to blindly believe or ignore them, it’s to extract their signal.

I’ve let pessimists talk me out of ideas that, looking back, were actually promising.
I’ve also had pessimists save me from spending years on problems where I had no shot of winning.
Both types exist. You have to learn how to tell them apart.

That one conversation was enough to validate Layout’s core value: turning a two-week manual local marketing workflow into a 2-minute automated process. From there, we had confidence to start building, knowing we were solving a real, high-impact problem.

With this approach, you don’t need months of discovery to at least commit to knowing more.
You can get to initial conviction, just enough to buy your 10th domain, vibe code a landing page, and start talking to actual customers, in a week or two.

And the second-order benefit:
It also gives you the foundation for your early sales motion.
(I might write about that in another post.)