How to Evaluate an AI Consulting Firm

If your Kenilworth firm is fielding pitches from AI consultants right now, here is what most of them have in common: they lead with a platform, a dashboard, or a process map. They cannot give you a fixed price. And six months later, your team is trained on a tool nobody uses, and the workflow that was costing you eight hours a week is still costing you eight hours a week.

The consultants who actually move the needle for professional services firms start somewhere different. They start with your P&L, not their stack. The seven questions below will help you tell them apart before you sign anything.

To evaluate an AI consulting firm: ask whether they start with your workflow or their product, demand a fixed-scope option, verify they have shipped production AI in the last 12 months, and confirm the deliverable is a working system you own, not a slide deck or a pilot that needs a second engagement to scale. A firm that hedges on any of those four points is selling something other than outcomes.

Why Most AI Consultants Sell the Wrong Thing

The AI consulting market is crowded with two kinds of firms. The first built something and decided to advise others on it. The second noticed AI was in every boardroom conversation and rebranded an existing practice.

Neither category is automatically wrong. But the second category, the strategy advisors who have never shipped a production AI system, tends to solve the wrong problem. They audit your operations, produce a roadmap, present it at an off-site, and leave you with a deck and a vendor recommendation. The question that actually matters for your firm's margin goes unanswered: what runs faster now than it did before we started?

Peter Drucker captured the management version of this clearly: "What gets measured gets managed." The same logic applies when hiring an AI consultant. If a firm cannot tell you exactly which metric changes after their engagement, they are not managing outcomes. They are managing optics.

The other structural problem is incentives. Many AI consulting firms earn referral fees from the platforms they recommend. That does not make their advice wrong, but it does mean their starting point is the product, not your firm. The real cost of an AI project is rarely the line item on the quote. It is the opportunity cost of six months spent on a project that changes nothing operational.

What Does a Real AI Engagement Actually Deliver?

A real AI engagement has a narrow scope, a clear deliverable, and a before-and-after you can measure in your own data. The deliverable is a working system, not a recommendation. When it is over, a workflow runs differently than it did before. That workflow either saves your team time, reduces errors, or eliminates a category of work that was slowing revenue.

Here is what that looks like in practice for a professional services firm on the North Shore:

• A working intake pipeline. Not a flowchart. An actual system where a prospect's inquiry is received, parsed, triaged, and routed without a human touching it by hand.
• A document extraction workflow. Not a proof-of-concept. A process where client statements, contracts, or disclosure documents are pulled into structured fields in your CRM automatically.
• A client communication draft engine. Your firm's voice, trained on your own templates, producing first drafts of follow-up letters, quarterly updates, or renewal summaries that your team edits rather than writes from scratch.

If a firm cannot show you a comparable before-and-after, they have not shipped one. That is relevant information. McKinsey's AI research consistently shows that the firms producing the most durable value from AI are the ones connecting it to specific business outcomes rather than treating it as a technology initiative. The breakdown of what AI consultants charge in 2026 gives you a realistic sense of what market-rate engagements look like by tier, and what to expect at each price point.

The 7 Questions to Ask Before Signing

These questions are not gotchas. They are diagnostic. A firm with real depth will answer them directly. A firm selling technology will hedge.

The comparison below maps how a technology vendor typically responds versus how an outcome-focused firm responds to the same set of criteria.

Our professional services AI practice is structured around fixed-scope projects: a defined workflow, a delivery window, and a system you own when we are done. That is the structure because it is the only one where the incentives point the same direction as your outcomes.

How Do You Know a Firm Has Real Technical Depth?

Technical depth is harder to audit than it sounds, because the vocabulary of AI is accessible to anyone who has read a few blog posts. A firm can talk fluently about large language models, retrieval-augmented generation, and multi-agent workflows without having built any of them.

Here is the faster test: ask them to describe the last system they built at the model level. Which model? Why that one for that task? How did they handle context length? Did they use Claude, GPT, or an open-weight model, and what drove that decision? A firm that has shipped a real system has specific, opinionated answers. A firm that reads about AI gives you a general answer that covers all options without committing to any of them.

You can also ask about team structure. Are the people advising you the same people who write the code? Some larger firms sell strategy through a senior consultant and deliver through a junior team in a different city. That is not inherently bad, but you should know the structure before you sign, not after the kickoff call.

One more test you can run yourself. Copy this prompt into Claude and use it to prepare for your next consulting conversation:

The quality of Claude's questions will mirror the quality of questions a real builder would ask. If your prospective consultant cannot answer them, that tells you something. Take our AI readiness quiz as a baseline for your own firm's current state before entering any consulting conversation. The more specific you are about your workflows, the harder it is for a vendor to sell you something you do not need.

Research published by Harvard Business Review on management consulting consistently shows that the gap between what consultants promise and what they deliver is widest when clients have no independent technical baseline of their own. An informed buyer is a protected buyer.

Red Flags That Should End the Conversation

Some signals do not require a seven-question diagnostic. They are visible in the first thirty minutes.

• "We will start with an AI readiness assessment." This is a fee for a diagnosis that should be part of scoping, not a separate deliverable. A real firm assesses readiness simultaneously with scoping, at no charge, as part of a discovery call.
• ROI projections with no cited source. "Our clients typically see a 40% reduction in admin time." Ask where that number comes from. If they cannot point to a specific client with a measurable before-and-after, treat it as a marketing claim, not a data point.
• No fixed-scope option. If every proposal is open-ended, the incentive structure favors the consultant, not you. That does not mean they are dishonest. It means your interests are not aligned, and that matters more than their intentions.
• They pitch broad transformation instead of a specific deliverable. Phrases like reshaping your firm's operations or unlocking AI potential signal a strategy play, not an implementation engagement. No Kenilworth professional services firm needs a transformation. They need a specific workflow to run faster, cheaper, or more reliably.
• They cannot name the model they would use for your project. If they say "we evaluate the best model for each use case" without naming which models are in their current toolkit and why, they are not builders. Builders have opinions. Advisors have frameworks.

The Bezos point applies directly here. An AI consultant who has never had an unhappy client has either never shipped at sufficient scale to generate real feedback, or is not being candid with you. A track record of real projects includes real failures and real learnings. Ask about them. The answer tells you more than any case study will.

The right AI consulting firm for your Kenilworth practice is not the one with the largest team or the most impressive logo wall. It is the one that can name the workflow they are solving, show you a comparable system they have shipped, and hand you something you own when the project ends.

If you want a direct conversation about where AI fits in your firm's operations, book a free 30-minute audit at Bace Agency. No pitch deck, no platform demo. Just a specific diagnosis of which workflow is costing you the most time right now.

Frequently Asked Questions