How to choose an AI consulting firm.

01

What are you hiring for?

Most companies come to this decision with a symptom, not a job description. The board has asked for a plan. A pilot has run for a year with nothing to show. A vendor has quoted something expensive and no one inside can judge it. Start by naming which of the three jobs you need.

1. A decision

   Which initiatives to fund, in what order, at what cost, and how you will know they worked. The output is a written verdict, not a deck of options.

2. A build

   One system taken from a working pilot into daily use, with a number that moved and a person on your team who can run it.

3. Ownership

   Someone senior to run the portfolio after the first system ships. The reviews, the build-or-buy calls, the report to the board.

The order matters. A build without the decision is how companies end up with a well-made tool nobody asked for. Ownership without a build is a retainer with nothing to own. If you cannot say which job you need, the first job is the decision.

02

What kinds of firms are there?

Three, broadly. Each is good at something the others are not.

1. The large consultancies

   McKinsey, BCG, Bain, Accenture, Deloitte, and their peers. They bring scale, industry benchmarks, and a name your board already knows. They can staff a fifty-person program across countries and regulated systems. The trade is that the partner who sells the work is rarely the person who does it, strategy and implementation often sit in different teams, and cost rises with headcount. They fit a program that is large, global, or heavily regulated.

2. The software builders

   Engineering agencies and product studios. They ship, on fixed scope, with strong engineers. The trade is that they build what you specify, so if the open question is what to build, you get the wrong thing built well. Adoption and measurement are usually outside the scope, and the team leaves when the code is delivered. They fit when you know exactly what you need and have someone inside to own it.

3. The boutiques

   Small partner-led firms, ours among them. The person who scopes the work does the work, and the firm runs a handful of engagements at a time. The trade is capacity. There is no bench, so a boutique cannot staff a large program, and you are hiring two or three specific people, whose track record is the whole story. They fit one decision or one system at a time, for a leadership team that wants the same people from the first call to production.

03

What should you ask before you sign?

Ask every firm the same six questions and compare the answers side by side. The answers separate the kinds of firm faster than any pitch.

1. Who will do the work?

   Names, not titles. Then ask whether those people will still be on the engagement in month three.

2. What do we walk away with?

   Ask for the artifact by name. A written decision. A system running in your environment. A measured number. If the answer is a framework or a roadmap, ask what a roadmap lets you do on Monday.

3. How is the fee set?

   Fixed per engagement, per person per day, or a monthly retainer. Each can be fair. What matters is that the fee and the scope are written down and agreed before the work starts, and that a change in scope means a conversation, not an invoice.

4. What happens the week after you leave?

   Who on our side owns the system. Who watches the number. What breaks if the person who built it is gone. A firm that has thought about this will answer in specifics.

5. How do you measure whether it worked?

   If the answer is adoption, engagement, or hours saved, press for the number your finance team would accept. Ask to see that number from a past engagement.

6. What do you not do?

   A firm that cannot answer this will take any work. The useful answer names the work they turn down and who they send it to.

04

What are the warning signs?

None of these is fatal on its own. Two or more together usually are.

• A proposal that describes the process in detail and the deliverable in one line.
• A pilot with no production date and no owner named on your side.
• The senior person who sold the work is not on the project plan.
• Case studies with no numbers, or numbers nobody will stand behind on a call.
• A price that only goes up as the scope becomes clear.
• Yes to everything. The firms worth hiring turn work down.

05

Where does a boutique like Fulkerson Advisors fit?

We do the three jobs above, one at a time, each led by a founding partner who stays through production. The Diagnosis is three weeks and ends in a written verdict. The Build is twelve weeks and ends with one system in daily use and a named owner on your team. The Standing Partner is two to four days a month for as long as you need someone to run the portfolio. The fee for each is agreed before we start.

Where we do not fit: a program that needs dozens of people, a board that wants a name it already knows on the cover, or work that is mostly systems integration across many countries. If that is you, hire one of the large consultancies and ask it the six questions above.

Ten case files are on the record, each with the challenge, what we built, and the outcome.

06

What should the first conversation look like?

Short, and specific to you. You describe the problem or the decision in front of you. The firm tells you which of the three jobs it is, whether they are the right people for it, and if they are not, who is. No slides. If a first call ends with a proposal and no diagnosis of your situation, you have learned something about the firm.

Ours is thirty minutes with a founding partner.