The AI hire you should not make

There is a job posting we keep seeing from companies between $2M and $30M in revenue. The title varies: Head of AI, AI Transformation Lead, Director of Innovation. The salary lands between $140K and $200K. The mandate is always the same fog: "define and execute our AI strategy."

Do not make this hire. Not yet. Here is what actually happens when you do.

The year you lose

Month one through three, your new Head of AI runs a listening tour. They interview department heads, map workflows, and build a slide deck about opportunity areas. The deck is usually good.

Month four through six, they evaluate vendors. Demos, trials, procurement calls. They discover what everyone discovers: the tools are generic, the integrations are shallow, and nothing fits your actual operation without engineering work.

Month seven through nine, they try to get engineering resources. But your engineers, if you have any, are busy running the business. So the Head of AI either fights for headcount or hires an agency, and now you are managing a manager managing a vendor.

Month ten through twelve, something small ships. A chatbot on the help page. A summarization tool three people use. The board asks what the AI strategy delivered, and the honest answer is a deck, a pilot, and a burn rate.

That is the standard arc. We have watched it from the inside more than once. Nobody in it is incompetent. The sequence itself is broken.

Strategy is downstream of systems

Here is the inversion that fixes it: you do not need an AI strategy to ship your first AI system. You need your first AI system to have any real AI strategy at all.

Until something is running in production against your actual data, on your actual workflows, with your actual team touching it daily, all AI planning is speculation. The first shipped system teaches you more in three weeks than a strategy hire learns in three quarters. What breaks. What your team trusts. Where the data is lying. Which workflow eats the most hours. You cannot slide-deck your way to any of that.

What to do instead

Pick the ugliest repeatable workflow in your business. The one where someone retypes data between systems, or reads the same kind of document over and over, or chases the same follow-ups every week.

Ship one system against it. Small, production, measured. Four to eight weeks, not four quarters.

Then look at what you learned and pick the second system. After three or four of these, a real strategy will be sitting in front of you, written by evidence instead of a consultant. That is the moment a senior AI operator hire makes sense, because now they have systems to run instead of a fog to define.

This sequencing is most of what people are actually buying when they work with us. Not code. Order of operations.

The audit is where we find that first ugly workflow. Ninety seconds, and considerably cheaper than a $180K listening tour.