I Quoted Three Weeks. It Took Seven Months.

Years ago I took a project to build an email application. I was excited, because it was the first big check. I told the client three weeks.

It took seven months.

I want to be precise about what went wrong, because the useful lesson is not the obvious one. The code was fine. My team was competent. Nobody got lazy. What happened was that I quoted a number I had not earned the right to quote.

We underestimated our own capacity. Not our skill, our capacity. There is a difference and it is the one that gets people. I knew what my team could build. I did not honestly account for what else they were already carrying, or what happens to a timeline when the person doing the integration is also doing three other things.

We underestimated the integration. Email looks simple from the outside. It is not. Every provider behaves slightly differently, the edge cases are endless, and the parts that break are the parts nobody demonstrates.

And I did not have the money to research it properly before I quoted. This is the part I think about most. The research would have cost a few thousand dollars and a week. Skipping it cost six months of margin and most of a client relationship. The research was the cheapest thing I skipped.

What I do differently now

I ask three questions before I quote anything.

What are my three biggest unknowns? If I cannot name them, I am not ready to price the work. Naming them is often uncomfortable, which is exactly why people skip it.

Who else is this team carrying? Capacity, not skill. Always capacity.

What would make me refuse this? I write it down before I want the money. Deciding in advance, while I am calm, is the only reliable defense against my own enthusiasm.

Why I tell this story from stage

I run a company that builds AI systems, and I speak to executive teams about adopting them. Almost every organization I meet is about to make a version of my mistake: committing to a timeline and a budget based on a demo and an instinct rather than on research they could have afforded.

When I tell an executive committee that their first AI project should be small enough that cancelling it is not embarrassing, I am not being cautious on their behalf. I am telling them what my seven months cost me.