How Do I Know If My AI Consultant Actually Knows What They're Doing?

The market rewards visibility over depth. The consultants with the loudest LinkedIn presence are not always the ones who can transfer capability to your team. Most buyers do not have an internal reference for what "good" looks like in an AI implementation, so the evaluation defaults to confidence, slide quality, and whether the person uses the right acronyms.

There is also a structural mismatch in most procurement conversations. You are trying to buy an outcome. The vendor is trying to sell an engagement shape they already know how to deliver. Without a disciplined interview, you will hear the shape, not the fit.

Finally, references are easy to game. "We helped a Fortune 500 company with AI" is not a reference. It is a category. A real reference names the workflow, the owner, and what happened after the consultant left.

Ask for the discovery story first. Before any solution content, ask how they would spend the first two weeks with your team. You want to hear: who they talk to, what they are trying to learn, and what artifact they expect to produce before a tool decision. If the answer is workshops and maturity assessments without individual contributor conversations, you are not evaluating an implementer.

Demand a single-workflow proof philosophy. Competent consultants anchor early work in one workflow, one owner, and one measurable outcome. If the proposal starts with a department-wide rollout or a "center of excellence" before anything works in one place, push back. Breadth is how engagements absorb budget without generating proof.

Separate demo skills from transfer skills. Watching someone prompt beautifully in a conference room is not the same skill as teaching your team to maintain the workflow. Ask explicitly how knowledge transfer is built into the engagement — what your people will be able to do without the consultant in the room.

Check for dependency patterns in past work. Ask what the client could still operate six months after the engagement ended. If the answer is mostly about deliverables and documentation rather than internal owners running live workflows, you are looking at a dependency model.

The interview question that surfaces the truth fastest is not technical. It is ownership.

Ask: "Who on our team will be able to adapt the first workflow without you?" If the consultant hesitates, qualifies heavily, or redirects to a support retainer, you have learned something valuable. The hesitation might be legitimate caution — but it might also be a business model built on staying indispensable.

CoCreate's evaluation conversations with prospective clients often start with a reverse discovery: we ask what they have already tried, where pilots stalled, and whether anyone internal can describe what was built. The gap between "we bought Copilot" and "here is the workflow, here is the owner, here is the before-and-after" tells you more than any vendor scorecard.

When a consultant can narrate a past engagement with those specifics — the bottleneck, the first workflow, the internal champion, the measurement method — you are in a different conversation than the one that starts with transformation language and ends with a statement of work.

If you want an external read on a proposal before you sign, CoCreate outlines how those conversations fit into its consulting services.

• Should I Hire an AI Consultant or Build the Capability In-House?
• Why Do Enterprise AI Pilots Fail?
• How Do You Use AI Without Putting Client Data at Risk?