A lot of people call themselves AI consultants now. Some of them are excellent. Some of them watched a few tutorials and added a line to their LinkedIn headline. The hard part for a small business owner is that both groups give roughly the same pitch, use the same words, and run demos that look impressive in the room.
So this guide skips the words and focuses on what you can actually check. If you are about to spend real money getting AI into your business, here is how to choose the person or firm who will earn it back, and how to spot the ones who won't.
Start with what you are actually buying
Before you evaluate anyone, get clear on what you want out of the engagement. You are not buying "AI." You are buying a specific outcome: hours recovered from a recurring task, faster turnaround on a process that bottlenecks your team, fewer errors in work that has to be right every time. A good consultant will push you toward that kind of specificity in the first conversation. A weak one will keep the conversation about the technology, because the technology is the only thing they are confident talking about.
That difference matters more than anything else in this guide. The right partner treats AI as a means to a result you can measure. The wrong one treats it as the product. Keep that lens on through every conversation that follows.
Four things to look for
These are the signals that separate a consultant who will produce a result from one who will produce a slide deck. None of them are about credentials or the logos on a website. They are about how the person thinks and how they work.
They ask what you will measure, not what you want to build
The first real question out of a good consultant's mouth is some version of "how will we know this worked?" They want a number attached to the project before any tool gets chosen, because that number is what tells both of you whether the engagement was worth it. If someone is ready to recommend a platform before they understand how you would measure its impact, they are selling, not advising.
Ask them: "Ninety days after we deploy this, what specific number should have moved?"
They understand your industry, not just the software
AI tools are mostly the same across industries. The workflows are not. A consultant who has worked in your world knows where the real friction sits, what data you actually have access to, and which "obvious" automation will blow up the first time it meets a real client. Generalists can still do good work, but they spend your money learning your business. Someone with vertical depth starts further down the field.
Ask them: "Walk me through a workflow like mine that you have automated before."
They hand you something you can run without them
The goal of a good engagement is not to make you permanently dependent on the consultant. It is to leave you with a system your team understands, documentation they can follow, and the ability to maintain and adjust it. Ask directly what you walk away owning. If the answer is a black box only they can touch, you have not bought a capability. You have bought a subscription to their availability.
Ask them: "When the project ends, what do my people actually own and operate?"
They will tell you when the answer is no
A consultant worth hiring will sometimes tell you that you are not ready, or that a cheaper off-the-shelf tool solves your problem without a project at all. That honesty costs them revenue in the moment and earns them trust that is worth far more. Someone who says yes to everything you propose is optimizing for the sale, not for your result.
Listen for: a recommendation that makes them less money but serves you better.
Red flags worth walking away from
Gartner has warned about "agent washing": vendors rebranding existing chatbots, AI assistants, and automation tools as agents without substantial agentic capability. In our experience, the same relabeling happens in consulting. These are the patterns that should make you slow down before you sign anything.
Watch for these
- The tool comes before the problem. They name a platform in the first meeting, before they understand what you are trying to fix.
- The pitch is "transformation," and nothing is measurable. Big words, no number anyone can check at 90 days.
- Pricing arrives with no scoping conversation. A real quote follows a real understanding of your work, not a rate card handed over on the first call.
- You cannot get a reference or see real work. Not a logo wall, an actual person you can call who paid them and got a result.
- Everything is proprietary. You are not allowed to understand how it works, which means you will never be able to maintain it.
How to compare engagement models
AI consultants package their work in a few common ways. None of them is automatically right. The best fit depends on how much certainty you have about the problem and how much ongoing change you expect.
Assessment first
A short, scoped engagement that maps where AI would actually pay off in your business before any building happens. This is the lowest-risk way to start, especially if you are not yet sure what to automate. You end up with a prioritized plan and a clear-eyed read on whether the bigger investment is worth it.
Project based
A defined scope, a defined deliverable, a defined price. This fits when you already know the workflow you want handled and the success criteria are clear. The risk to avoid is a fixed price wrapped around a fuzzy goal, which is how projects drift and get abandoned.
Retainer or managed
Ongoing work where the consultant builds, monitors, and optimizes over time. This makes sense when AI is becoming part of how you operate and you want someone accountable for keeping it working as your business changes. Make sure the retainer is tied to outcomes you review regularly, not just hours logged.
Fractional or advisory
Senior guidance on a part-time basis, steering your internal team rather than doing all the hands-on work. This fits if you have capable people who need direction more than they need labor. It keeps institutional knowledge inside your company.
Whatever the model, the contract should make the exit obvious. You should always know what you own, what it costs to continue, and what happens if you stop. Clarity on the way out is the best sign of an honest engagement on the way in.
Not sure what you should be hiring for yet?
Start with the part that should come first. The QP AI readiness assessment maps where AI would actually pay off in your business, what you would need in place to deploy it, and what a realistic result looks like, before you commit to any project or any consultant.
See the AI readiness assessmentThe one test that cuts through everything
If you only remember one thing from this guide, make it this. By the end of your first real conversation with an AI consultant, you should be able to write down a single sentence: the specific process they will improve, and the number that will tell you it worked.
If you cannot finish that sentence after the call, it is not necessarily a bad consultant. It might just mean you have not done the groundwork to know what you are buying yet. Either way, you have your answer about what to do next, and you have saved yourself from signing a contract built on a hope instead of a plan.
Choosing well is not about finding the consultant with the most impressive demo. It is about finding the one who insists on the same discipline you would apply to any other serious decision in your business. Define the problem. Define success. Build to that. Measure the result. The right partner will hold you to that standard, because it is the only standard that produces something worth paying for.
If you want to talk it through, we are easy to reach, and the AI readiness assessment is a low-risk place to start. QP works with businesses across Massachusetts.