The One Path Decision

When I started BidPilot on June 8, I gave it a very large job title.

An autonomous construction revenue platform.

Even typing that now makes me laugh a little. I am a 50-year-old non-technical person using AI to build software, and here I was aiming at something that sounds like it belongs in a glass conference room with a team of people in fleece vests.

But the idea mattered to me.

Construction suppliers, steel building dealers, and subcontractors deal with a lot of back-and-forth. Quotes come in. Details have to be pulled out. Materials need to become a bill of materials. Proposals have to be written. Follow-ups need to happen. If the work turns into a real job, it has to become a project.

That chain was the heart of the product:

quote upload → AI extraction → BOM → proposal → project

And the design decision I obsessed over was whether to build the whole giant machine in my head, or just build one clean path through it.

The phrase for that is “MVP happy path only.” I know, another phrase. MVP just means the smallest useful version. Happy path means the simplest normal route, where nothing weird happens.

In real life, of course, construction is mostly weird things happening.

A quote might be incomplete. A line item might be vague. A supplier might describe the same part three different ways. A customer might change their mind. A bid might need special notes. Someone may upload something that barely looks like a quote at all.

My instinct was to design for all of that right away.

I wanted the app to be smart enough to handle every exception. I wanted it to feel serious. I wanted it to look like it understood the industry. And, if I’m being honest, I also wanted to protect myself from the little voice saying, “Who are you to build this?”

So I kept circling around edge cases.

What if the AI pulls the wrong number? What if the proposal needs three versions? What if a dealer wants a different workflow than a subcontractor? What if steel buildings need one structure and general suppliers need another?

Those were real questions. But they were also quicksand.

The more I tried to design around every possibility, the less the product had a shape. It became a foggy beast with a thousand arms. I could describe the ambition, but I could not tell you what the first user would actually do on screen from minute one to minute five.

That bothered me.

Not in a fancy product-strategy way. In a plain old “I am lost” way.

So I made the decision: one path first.

Upload a quote. Let the AI extract the useful details. Turn those details into a bill of materials. Use that to generate proposal copy. Then create a project from the accepted work.

No branching universe. No giant settings maze. No pretending I had already solved every construction workflow on earth.

This mattered to me because I needed the app to be understandable before it could be powerful.

If I, the person building it, could not follow the path, then a busy supplier or subcontractor definitely would not want to follow it. People in construction are not sitting around hoping for more software to babysit. If anything, they are probably suspicious of it, and I don’t blame them.

The app had to earn its keep quickly.

That meant the design had to say, without a long explanation: “Give me the quote. I’ll help you turn it into the next useful thing.”

There is still a massive scope behind BidPilot. The bigger ambition has not gone away. Follow-ups, project creation, a whole revenue lifecycle — that is still the map.

But I had to stop trying to build the whole map at once.

I had to build a road.

And as simple as that sounds, it was not easy for me. Saying “happy path only” felt like admitting the first version would be incomplete. Which it is. But incomplete is not the same as unserious.

Sometimes it is the only honest way to begin.

Small takeaway: a clear first path beats a perfect imaginary system.