An AI consultant does the AI work your organization cannot yet do for itself. In practice that means teaching your people to use AI well and building the automations and agents your team will run. It also means advising the leaders who decide where AI belongs. Everything else on a consultant’s website is a version of one of those.
The title stays vague because the work refuses to sit still. Some weeks the job is teaching prompt technique to a room of marketers. Other weeks it is telling a leadership team that the pilot everyone loves should be shut down. Same job, one title. What follows is the job as we actually run it, so you know what you are evaluating before you evaluate anyone.
What does the work cover?
Start with training. Most organizations do. Our flagship is a hands-on bootcamp: teams pick a build from a preset catalog of seven, then take that one system end to end in their own AI accounts. No demo environments. Fast teams get through two. Around the flagship sit role-based workshops tuned to specific functions, and keynotes for rooms that want the argument before they need the skills.
The second family is builds: workflow automation and AI agents, scoped as projects with a defined finish line. A claims team might want first-pass summaries of incoming files. An operations team might want an assistant that fields the questions the wiki already answers, in the channel where people ask them. Either way, the deliverable is a working system your team operates.
The third family is advisory. Someone has to own the AI roadmap and answer for its results, and most organizations do not have that person on payroll yet. A fractional Chief AI Officer fills that seat on a retainer without an executive search: strategy, vendor calls, governance, and a monthly cadence that keeps the roadmap moving.
One service sits apart from the families: AEO, answer engine optimization, which makes your content legible to AI search. We sell it and we practice it. The spec sheets and FAQ schema on this page use it on our own content.
When do organizations bring one in?
Later than they meant to, usually. By the time an organization calls, it has already tried. Licenses were bought. A lunch-and-learn happened. One or two people got quietly good, and everyone else started routing work to them. Results are real and anecdotal at the same time, and nobody owns the outcome.
The moments that turn into engagements look like this:
- Licenses are paid for and lightly used.
- One enthusiast carries every AI question in the building.
- A pilot worked and nobody can say what it changed in numbers.
- The board wants an AI plan and the current plan is a slide.
- Something customer-facing is about to ship with a model inside, and legal has questions.
All five point the same way. The tools are already in the building. What is usually missing is anyone whose job is to change how the work gets done around them.
What does good look like?
Good consulting leaves your people holding the tools. We hold ourselves to that test too. Our bootcamp builds happen in the team’s own AI accounts on purpose: when the engagement ends, the system and the skills stay where they were built. A consultant whose value walks out the door with them rented you a demo.
It should also have a number attached. Before work starts, agree on what will be different and who is counting. We scope around measured ROI, so there is a result to point to once the initial enthusiasm wears off.
Expect good consulting to turn work down. A no or two on the first call is a good sign: some processes are not ready for automation, and some pilots deserve an early shutdown before more money goes in.
Last, listen to how a consultant talks about your people. Rollouts succeed or stall with the people being asked to change how they work, and they can usually tell whether a plan accounts for them.
Backgrounds are worth one question too. Ours: the practice is run by Dexter Brocks, the Chicago AI Guy, self-taught from the support desk to cloud engineer. Support desks teach a durable habit, which is fixing the thing the person in front of you is actually stuck on. He teaches in public as well, including at the University of Chicago’s Polsky Center.
How does an engagement run?
It starts with a 30-minute call, free, and we put numbers on it before it ends. We do that because vagueness at the pricing stage tends to continue at every stage after it.
Training engagements are the most structured. The agenda is set in advance and teams pick their builds from the preset catalog. Bootcamps include 30 days of support after the room empties. Workshops can add the same 30 days as a paid add-on.
Builds are quieter and longer. We agree on what finished means, then build in your systems, and the handover leaves a working system your team runs without us on the phone. Automation and AEO builds include support after delivery, because a system nobody supports after handover quietly stops working.
Advisory never really hands over. The fractional seat runs on a monthly cadence for as long as it earns its keep, and part of the work is getting the organization ready to need us less.
Geography is the simple part. Based in Chicago, traveling wherever your team is. Training happens where your people sit. Build and advisory work runs in your systems, wherever those live.
What does it cost?
The shape of the price is public even when the number is not. Two of the three settle on a number before anything starts: training carries a fixed scope and a fixed price, and a build is priced by the project. Advisory is the standing exception. It runs on a monthly retainer, because the seat has no natural end date.
Published numbers would be guesses, because scope moves them. So the numbers come on the first call. The call is 30 minutes and it is free. Bring the workflow that eats the most of your team’s attention. You will leave with a number and a recommendation, and both are yours to keep either way.