Price Tag

A story comes across the feed. A young company built a project quoted at tens of thousands of dollars in a few days using AI coding. Did it in-house.

I get it. I had that phase too. I'd look at an SI vendor's estimate and think, how does it cost that much. I could build it faster and cheaper. They must be padding the bill because they lack the skill. In my twenties, I genuinely believed that.

Years in the industry teach you to read between the lines of an estimate.

A contract between companies is not a promise to deliver code. It's a promise backed by your existence. You nail down requirements, design, implement, test, deliver, operate, and when something breaks, you fix it within the bounds of the contract. It's not just warranty clauses. The company's name and credibility hang from every line item. Cut corners and word gets around. The next deal doesn't come. So an estimate in the tens of thousands isn't just paying for code. It includes the insurance premium against reputational risk, and the result of cold math on future revenue.

A prototype someone built over a weekend and a system a company signs off on and delivers — they may be written in the same language, but they're different creatures. If the first one breaks, you fix it. If the second one breaks, damages and loss of trust come as a package.

AI made writing code dramatically faster. That's a fact. But most of an estimate isn't composed of time spent writing code. Consensus building, wrestling with ambiguous requirements, testing, operational design, and the promise that you won't run when something goes wrong. None of that comes from a prompt.

Fearless young people, armed with AI, can now do bigger things in less time than ever before. That in itself is wonderful. But the real fear starts after the accident. And that kind of fear can't even be imagined until it happens.