Where Does AI Fit? Write the Job Description First

"Where does AI fit in our business?" It is the question I hear most as an AI advisor, and almost everyone asks it the wrong way around. They start with the technology and go hunting for a home for it. After 35+ years in software and security, and as the working CTO of my own companies, I use a much simpler filter, and it comes from hiring, not from tech.

If you cannot write the job description, you should not make the hire. The same rule applies to AI.

The hiring test

Think about the discipline you already apply to bringing a person on board. You define the role. You list what they receive, what they produce, and what success looks like in ninety days. You decide who they report to and who reviews their work. Nobody signs an offer letter for a role called "transform the company."

Yet that is precisely the role AI keeps getting handed. A mission statement instead of a job description. "Make us AI-first." "Modernize our operations." Those are slogans, and slogans do not ship.

So before any AI project gets a dollar from me, it has to pass four questions. What are the inputs it receives? What output does it owe? What does done look like? Who checks the work? Answer all four and you have a real AI opportunity. Miss even one and you do not have an AI project yet. You have a wish with a budget attached.

Constraint is where the value hides

A story from my own resume. At a sports video startup, we built AI to find marketable moments in football footage using facial recognition. Pointed at every player in the Premier League, the system produced garbage. It once tagged a player who had been dead for twenty years. The fix was not more AI, a bigger model, or a larger budget. We constrained the problem to the twenty-two players actually on the pitch, and accuracy soared.

That is the job description principle in action. The AI did not fail because it was weak. It failed because we had given it a vague role. The moment we scoped the job tightly, the same technology became genuinely valuable. Every experienced operator I know has a version of this story, and every disappointed AI buyer I meet is missing one.

Scope it like a probation period

The second half of the hiring test is what happens after day one. A new employee gets onboarding, a manager, and a probation period. Their responsibilities grow as trust grows. AI deserves exactly the same treatment, and almost never gets it.

Give AI one task. Make it repetitive, high volume, and cheap to get wrong while you are learning. Measure the output the way you would measure a new hire on probation: sampled, reviewed, and owned by a named person. When it earns trust, expand the role. When it underperforms, pull the scope back without drama, the way a good manager would.

And insist on the management question up front. Who reviews this system's output? Who owns its mistakes? Who decides when it gets promoted into more responsibility? If nobody in the building manages the AI, the AI does not fit. Not yet. In my experience the technology is rarely the blocker. The missing manager is.

What this means for your next AI decision

This is where an outside perspective earns its keep. A good AI consultant or fractional CTO is not there to sell you more AI. The most valuable work is often narrowing the role: taking the grand mandate your team wrote and turning it into two or three tightly scoped job descriptions with clear owners, clear checks, and a measurable definition of done.

That reframe also answers the budget question honestly. You would never approve headcount without a role definition, a manager, and a review cycle. Hold AI spend to the same standard and the "where does AI fit" question mostly answers itself. The tasks with clean inputs, clear outputs, and a named reviewer rise to the top. The vague transformation projects, the ones that quietly die at ninety percent, never make the list in the first place.

AI is not magic and it is not a miracle hire. It is capable, tireless, and completely indifferent to your intentions. Scope it like a role, manage it like a new employee, and it will earn its place on the org chart faster than any hype cycle promised.

If you want a working CTO's eyes on the job descriptions AI should be applying for in your business, I take a few clients at a time. Book your free 15-min AI Fit Call →