How to Choose an AI Consulting Firm (and Avoid the PowerPoint Consultants)

The best AI consulting firm for your business is the one that diagnoses before it sells, has actually delivered in production, and leaves you with something your own team can run after they go. The worst is the one that arrives with a slide deck, a transformation roadmap and a plan to stay forever. Telling them apart before you sign is the most valuable skill a buyer of AI consulting can have. MIT's 2025 study found around 95 per cent of generative AI pilots delivered no measurable value, and a consultant who reinforces the usual mistakes will charge you well for the privilege.

What does a good AI consulting firm actually do?

A good firm starts by understanding your problem, not by presenting its capabilities. It diagnoses whether AI is even the right answer, and is willing to tell you when it is not. It scopes a specific outcome, prepares the data and process that outcome depends on, delivers something that works in production, and hands it over so your team can run and extend it. A weak firm does the opposite: it sells a transformation programme before understanding the problem, staffs the work with juniors after the senior team has won the pitch, builds something that only it can maintain, and arranges to be needed indefinitely. The difference is visible early, if you know what to ask.

The questions to ask before you hire an AI consultant

These questions separate practitioners from presenters — ask them directly and listen for specifics rather than adjectives. Have you delivered this in production, and can you describe exactly what you built? What would make you tell a client not to do an AI project? Who, specifically, will do the work day to day, and what is their background? What will my team be able to operate themselves after you leave? How do you measure whether the engagement worked? What does the data and process work look like before any AI is involved? A firm that answers these in concrete terms has done the work. A firm that retreats into language about ecosystems, leverage and transformation is telling you something useful by failing to answer.

Red flags when choosing an AI consultancy

Some signals reliably predict a disappointing engagement. The pitch leads with technology and ambition, not with your problem. The senior people who win the work disappear after signing, replaced by juniors. The plan is a multi-year transformation roadmap rather than a scoped, measurable outcome. The proposal contains no honest data or process work, as if AI sits on nothing. The deliverable creates a permanent dependency on the firm. And the ratio of buzzwords to specifics is high — which is one of the most reliable incompetence detectors there is. None is automatically disqualifying, but several together should stop you.

How much should AI consulting cost, and in what shape?

Be more interested in the shape of the engagement than the headline rate. The most useful AI consulting tends to come in short, scoped engagements with a defined deliverable, not open-ended retainers that bill by presence. A firm confident in its work will scope a diagnosis or a delivery with a clear end, because it expects to earn the next piece of work by being good, not by being embedded. Treat a proposal for permanent presence with caution. The aim of a good engagement is to make you more capable, not more dependent. A consultancy whose commercial model relies on you never becoming self-sufficient is not aligned with your interest.