Internal team or implementer for the first AI project?

For the first project, hire an implementer and use the project to form the internal team that takes over the operation afterward. Building a data team before you have the first criterion is hiring for a target that does not exist yet.

The logic is sequence. The first project defines what your operation actually needs from AI, and that is almost never what it seemed at the start. Whoever hires a senior team before that pays expensive salaries while discovering the problem, and discovers it slowly, because the new team does not yet know the operation. The right implementer works from inside, hands back a running project, and leaves the runbook and the criterion for your people to hold. The internal team keeps the operation at the end, formed by the project itself.

The exception is real and specific. If you already have an idle senior data team, with people who know the operation and have room on their calendar, building internally makes sense, and the implementer becomes a luxury. But that is rare. In most chains, the existing team is busy maintaining what already runs, and the first AI project becomes the project that never starts, pushed to the month that never comes.

What you can do this week is write, in one sentence, the success criterion for the first project. Not the technology, the result: shelf-out at this, margin at that, response time at this. If you cannot write it, the problem is not who to hire; it is that the project has no criterion yet. And most AI projects fail on ROI precisely because they start without one.

If you are in this decision and want a second read, send me the shape of your case in one sentence, the problem and the team you already have. I will reply within a business day with a paragraph: internal, implementer, or “depends, here is the missing question”. No calendar, no call.