Sense with Cents

The Polite No was about what happened when I went the usual route — local specialists, referrals, websites — and got a string of polite nos in return. After that path ran out, I tried a different route. I posted the same work on an expert-search platform, where practitioners bid on stated jobs and you pick from who responds. The polite no was no longer the problem. A different failure mode was.

Before posting, I used AI to format a draft of what I already knew I wanted. I dumped in our desires — what the document needed to do, who it needed to cover, the specific things we needed to protect — and AI organized it into a clean, logical layout. Then I told the tool to fact-check its own output against reliable public examples and flag anything that did not line up. That last step is the one some may miss, and probably the most important step in any AI workflow. It is what turns a confident-sounding draft into a draft you can stand behind.

That is the cleanup-tool use case, not the thinking-tool use case. The thinking was mine. The formatting was the tool's. The verification was the tool checking itself against sources I trust.

I had also looked at templates before I went the AI route. The templates I found were either way more than I needed or tilted in a direction that did not fit the specific situation. So I worked out what I actually wanted and used AI to lay it out — and then to check itself against the public record.

What I was looking for was a redline — a safety pass by someone qualified to catch what I might have missed. I disclosed up front on the platform how I had built the draft, including that I used AI. Full information at the top is the kind of thing I wrote about in Help Me Help You — give the professional everything they need to answer the actual question, and the answer comes back faster and better. It also doubles as a filter. Practitioners who responded knew exactly what they were responding to.

Several practitioners responded. I narrowed the field to two and replied to both. One was a 40-year veteran. The other was newer to the field, young enough that I wondered, briefly, if I was making a mistake going that direction at all.

The two responses

The newer expert wrote back: send the draft over, happy to take a look. Then asked a clarifying question that showed real reading — was I looking for one document covering the general use case, or did I need a separate one for each engagement. A small question, asked in one sentence, that surfaced something I had not fully thought through. We were on the substance inside of two exchanges.

The 40-year veteran wrote back something different. The reply did not engage with my draft at all. It said, in effect, that a dozen prior AI-prepared documents had come through the veteran's practice and all of them were full of mistakes — that it would cost more to fix one than to use a time-tested standard form. The offer was to take my details and run them through the veteran's own template.

I wrote back to explain. The templates I had found did not fit our specific situation. I had used AI to organize the protections I actually needed, not to invent them.

The reply was a one-liner asking whether I would rather trust software that hallucinates or a practitioner with 40 years of experience.

I hired the newer one.

One detail worth mentioning. Both quoted the same price. Neither was the lowest bid on the board. So this was not a story about the newer expert being cheaper, or about the veteran charging a premium-veteran rate. The platform likely nudges bids toward a suggested band, and it looks like both took the suggestion. The money was a wash. The choice was entirely about how each one handled the inquiry.

I should mention something

I have 40 years in my own field. And I was reluctant to use AI until the tool matured. I held off until I was nearly forced into it — customers and the work itself pulling me in faster than I would have chosen on my own. So when I describe the veteran's reaction, I am not describing some other type of person. I am describing the same kind of practitioner I am.

That is part of why the response landed the way it did. I recognized it. I have made the same mistake. I try to learn from experience, and the only reason I can write about this honestly is that I have been on the other side of it.

The actual failure

The veteran's pattern recognition probably has a basis. A dozen AI-prepared documents that all needed substantial rework is real operator experience. I respect that kind of experience. It is the same kind I run my own work on. After enough repetitions of the same problem walking through the door, you start to recognize the shape of it before the inquiry finishes describing itself.

That recognition is useful. It is also dangerous. I know this because I have done it. I have filed a new inquiry into the pattern from prior ones and answered the pattern instead of the question. Sometimes I caught it in time. Sometimes I did not, and the prospect had to push back to get me to actually look. Those are the inquiries I remember, because they were the ones where I had the experience to know better and let the experience get in front of the listening.

That is what happened with the veteran. The pattern from prior clients had a verdict ready. When I explained that the specific situation did not fit the off-the-shelf template — that I had already looked at templates and they did not fit, and that I had built verification into the AI step — the reply was a slogan, not a response to the explanation. Refusal to look at the specific case because the pattern already had a verdict ready. The trigger changes over the years. Twenty years ago it was self-diagnosed mechanical complaints. Thirty years ago it was customers showing up with printouts from a forum. AI is just the current shape of it. The failure mode is the same.

This is also where the buyer-seat view connects to something I wrote about from the seller's seat in AI Is a Power Tool. The Expert Still Has to Hold It. The experts who stay valuable are the ones who wield the tool — and, just as important, the ones who can engage with customers who wield it.

What I took from it

I run a small software company. People reach out before they ever become customers — asking whether the software fits their situation, what features it has, whether we handle their state. After enough of those inquiries, the questions start to look familiar. The temptation, when a new inquiry arrives, is to file it into the familiar pattern and answer the pattern instead of the question.

That temptation is a trap, and I know it is a trap because I have fallen into it. The prospect is in front of you. They have done some work. They are asking a specific thing. The job is to answer the specific thing — not to file them into the bucket they look like they belong in.

Challenging a prospect's workflow is bad business. It tells them that their judgment is the problem, before you have even understood what they were trying to do. Sometimes the workflow really is the problem and it needs to be said. But it is the second thing said, not the first. The first thing is hearing the question.

The 40-year veteran lost a paying customer by answering a question I did not ask. The newer expert got the job by answering the one I did. That is the lesson I am keeping — the one I have to keep relearning. Hear the question first. Everything else comes after.

Cleaned up with AI assistance. The judgment, the story, and the choice of practitioner are mine.