What This Is Really Costing

When I started MyUPP on June 6, I knew it was a big idea.

A national physician credentialing platform is not exactly a weekend recipe app. The dream is to help cut credentialing from the usual 90 to 180 days down to something closer to days. That means a place where a provider can build a reusable passport, and hospitals or payers can review what they need without everyone starting from scratch every time.

Even typing that out makes me laugh a little.

I am 50. I do not have a coding background. And somehow I am trying to build something with provider portals, hospital portals, payer portals, credentialing specialist tools, sign-ins, permissions, and a real database underneath it.

So yes, there is a cost.

The first cost is the obvious one: AI credits.

Every time I ask the builder to create a page, fix a mistake, connect something to the database, or rethink a flow, I am spending credits. Sometimes that feels completely worth it. I can describe what I want in normal language and watch something appear that I could never have built by myself from scratch.

Other times, I burn credits because I was not clear enough. Or because the app got too ambitious too fast. Or because I asked for five things at once when I should have asked for one small step.

That has been one of the first money lessons: vague thinking has a cost.

If I say, “Make this better,” the AI may do something. But if I have not decided what “better” means, I am paying for confusion.

The second cost is time.

Not coding time, exactly. Thinking time.

Before this project, I assumed building software was mostly about typing complicated computer language. Now I see that a lot of it is deciding. What belongs in version one? What has to wait? What does the provider see first? What should a hospital user be able to access? What is too much for now?

My original idea was massive. Provider, Hospital, Payer, and Credentialing Specialist portals. Document scanning later. A marketing landing page later. A visual style that felt like “Clinical Trust” with deep navy and medical teal. All of that sounds neat when written in a plan.

Inside the actual app, it can become a swamp.

So I had to narrow version one to the Provider Passport and the Credentialing Engine. Even that is still big. But at least it gives the work a spine.

The third cost is attention, and this may be the sneakiest one.

This project sits in my head even when I am not working on it. I find myself thinking about whether the database is set up correctly, whether permissions are safe, whether the sign-in flow makes sense, whether a non-technical provider would understand what to do next.

There is also the emotional cost of realizing that “the AI can build it” does not mean “the AI can decide it for me.”

That was a little disappointing, honestly.

I wanted the magic machine to carry more of the burden. And it does carry a lot. But I still have to be the adult in the room. I have to notice when the scope is too wide. I have to stop and ask whether this feature belongs now or later. I have to learn enough words like “RLS” — which is basically rules for who can see what in the database — so I do not accidentally build something careless.

The real cost of building MyUPP is not just credits on a screen.

It is learning to spend those credits with more care. It is accepting that slow thinking can save fast waste. It is resisting the urge to build the whole dream in one week.

I still believe in the idea. But I am learning that belief is not a budget.

Small takeaway: AI can make building faster, but it still charges you for unclear thinking.