I built two websites before talking to users. I think I see the problem now.

I've done the same thing twice now.

I get an idea.

AI makes it easy enough to build.

So I build the website first and start looking for the value afterwards.

The first was HeadSpaDirectory.com.

I thought a directory helping people discover head spas could be useful, so I quickly built the structure, added locations and listings, and started working on the site.

Only later did I start talking to actual spa owners.

Those conversations changed how I saw the project.

The more interesting problem wasn't simply:

"How do people find a head spa?"

It might be trust.

How do customers know whether a spa is actually trained in authentic Japanese head spa techniques?

That pushed me toward certification, practitioner background, and verified information things I hadn't understood when I built the first version.

Then I did almost exactly the same thing again.

I started AIStartupOfOne.com, a directory of AI-native one-person companies.

Again:

Idea → website → database → then conversations.

I've now started talking with founders.

And again, the conversations are changing what I think is valuable.

The founder profiles are useful.

But I'm becoming much more interested in the decisions behind them:

How did they decide what NOT to build?

What evidence was enough to keep going?

How did they get the first paying customer?

When does a solo founder stop building and start validating?

Here's the embarrassing part:

I already knew I was supposed to validate first.

My husband questioned the value and monetization of both ideas when I discussed them with him.

AI had even helped me map out the likely failure points of my first project.

None of this was new information.

And I still did it again.

That's what I'm trying to understand.

Maybe AI hasn't just made building easier.

It has made building feel like progress.

Talking to users is uncomfortable.

You send messages and get no reply.

People question your assumptions.

Sometimes you don't even know what question you should be asking.

Building is different.

You give AI an instruction.

Something happens.

A page appears.

A feature works.

You feel like you've moved forward.

Today I read a post here from a founder who shipped an AI job-hunting tool in three weeks but hasn't found paying users yet.

One line stood out:

"I can ship, but I don't naturally think about distribution. I built first and figured I'd tell people later."

I've basically done the same thing. Twice.

And it made me wonder if AI has created a strange new founder problem.

Building used to be a constraint.

Now one person can turn an assumption into a working product incredibly quickly.

That's amazing.

But maybe the dangerous sequence is becoming:

Idea → Build → Launch → Now let's find out if anyone actually needs it

instead of:

Problem → Conversations → Evidence → Build

The advice hasn't changed.

"Talk to users."

"Validate before building."

We've all heard it.

But maybe knowing the advice isn't the hard part anymore.

The hard part is resisting the temptation to build when building has become the easiest part.

I'm starting to think the scarce skill for solo founders is shifting from:

"Can I build this?"

to:

"Should I build this yet?"

I came across a line recently that stuck with me:

"Stories from your own business that nobody else has the receipts to tell."

Maybe this is one of mine.

Two projects. Same mistake.

I'm curious whether other founders are experiencing the same thing:

Has AI made you better at testing ideas — or mostly faster at turning assumptions into products?