Where Does AI Fit? Start With the Task Your Team Dreads

Every company I talk to wants to know where AI fits. Most of them look for the answer in the wrong room.

They look in the boardroom, or at the strategy offsite, or in a vendor's slide deck. The question gets framed as big and visible: "How do we use AI to transform the customer experience?" It sounds ambitious. It rarely ships.

The AI wins I have been part of, as a fractional CTO, as an AI advisor to clients, and as the working CTO of my own companies, almost all started somewhere much less glamorous: with the task the team quietly dreads.

A listing process that went from weeks to minutes

One of my favorite results came from a real estate company. Their new property-listing process had become a multi-week slog. Every listing crawled through the same slow, repetitive steps before anyone could actually sell it.

The tempting move was to aim AI at something impressive: marketing copy, lead scoring, a chatbot on the website. Instead we pointed it at the grind in the middle, the part of the process nobody was proud of and nobody would ever put in a strategy deck.

That process went from weeks to minutes. The AI itself was the least interesting part of the story. The interesting part was the choice of target. And the team got those hours back for the work customers actually notice.

Why dreaded work makes the best first AI project

When someone asks me where AI fits in their business, I spend the first hour away from leadership. I find the person who does the work and ask one question: what part of your week do you dread?

The answers are never exciting. Retyping the same details into three systems. Chasing the same approvals every Thursday. Rebuilding a report every Monday that nobody reads past page one. But that list is almost always the best AI shortlist in the building, for three reasons.

• It repeats. Dreaded work is usually dreaded because it comes back every day or every week. Repetition is where automation pays, because every hour you save multiplies.
• It is well understood. The person doing it knows exactly what "done right" looks like. That means you can check the AI's output quickly, which is the difference between a tool that saves time and one that just moves the work onto someone else's plate.
• Nobody is attached to it. Adoption is the quiet killer of AI projects. When you take away a task people hate, nobody fights you. When you automate the part of the job they enjoy, they route around your shiny new tool and keep doing it the old way.

The process you think you have vs. the one you actually have

There is a second reason to start with the person doing the job. Executives describe the process they think they have. The front line describes the one you actually have.

The org chart version of a workflow is clean: a request comes in, gets approved, gets done. The real version has a spreadsheet someone keeps on their desktop, a phone call that happens every single time, and a workaround nobody wrote down because it "just works."

If you automate the clean version, you automate a process that does not exist. The fix is cheap: sit with the person doing the work, watch it end to end, and write down every step they would never think to mention. It takes an afternoon. It saves a quarter.

What 35 years taught me about where AI fits

I have spent more than 35 years in software and security, including time as Senior Director of Technology at an AI startup, where I was employee #2. Every technology wave I have lived through, from the early web to mobile to cloud to AI, has produced the same pattern. The companies that get real value do not start with the technology. They start with a real, specific, slightly boring problem, and they let the technology earn its way in.

AI is genuinely powerful. It can take a multi-week slog down to minutes. But it is not the whole solution, and it will not find its own best use. Someone has to point it at the right task, and the people best placed to tell you which task that is are usually not in the boardroom.

A simple way to start this week

1. Ask three people who do hands-on work in your business: "What part of your week do you dread?"
2. For each answer, estimate how many hours a month it eats and how easy it would be to check if a machine did it.
3. Pick the one that is high in hours, easy to check, and low in risk if it goes wrong. That is your first AI project.
4. Before you build anything, watch that task done start to finish and write down every step, including the ones nobody mentions.

That is not an AI consultant's pitch for a big transformation program. It is the same judgment you already use when you decide what to delegate to a new hire.

If you want a fractional CTO and AI advisor to help you find the dreaded task in your business that is worth automating first, and to tell you honestly which ones are not, I take a few clients at a time. Book your free 15-min AI Fit Call →