The four kinds of firm.
"AI transformation consulting" is sold by firms whose work has very little in common. Naming the four shapes makes the buying decision shorter. McKinsey's State of AI research and BCG's AI value research both find that the firms reporting the highest AI ROI are not the ones spending the most on transformation consulting - they are the ones matching the firm shape to the actual job.
01 - Enterprise SIs
Global system integrators.
Accenture Applied Intelligence, Deloitte AI, IBM Consulting, Capgemini. Built for enterprise global delivery. Best when you have $1M+ to spend, multiple geographies, and need build plus run.
Watch-outs: overhead is real. Mid-market briefs get a junior team and a senior partner deck. See alternatives to Accenture Applied Intelligence.
02 - Strategy houses
McKinsey, BCG, Bain - and their AI arms.
McKinsey QuantumBlack, BCG X, Bain Vector. Best for boardroom-level AI strategy, portfolio decisions, and large-scale value cases. Strong on the strategy, lighter on the day-to-day adoption work.
Compare directly: vs McKinsey QuantumBlack, vs BCG X.
03 - Specialist AI shops
Faculty, Fractal, Quantiphi, product studios.
Best for AI builds, models, and product development. Not change firms. Pair them with a change partner when the brief is also about adoption.
04 - Change and adoption specialists
Prosci, Kotter - and AI-native change firms.
Classical change firms (Prosci, Kotter, McChrystal Group) are excellent on the human side but not AI-native. AI-native change firms - Average Robot among them - are built around the AI-at-work problem specifically. Best for mid-market teams whose strategy is broadly clear but adoption has stalled. Harvard Business Review's writing on making generative AI work repeatedly points to adoption, not strategy, as the binding constraint for most organizations.
How to brief the work.
- Name the gap in one sentence. Strategy, build, adoption, governance - pick one as primary. If you cannot, that is the first conversation.
- State the outcome in business terms. Not "ship the AI program". The number that should move and by when.
- Set a budget band. Firms shape proposals to budgets. A range avoids wasting cycles with the wrong tier.
- Ask for the team you will actually get. Not the pitch team. Named partners, named senior consultants, and the day-to-day operator.
- Require a written deliverable. Slides alone are not a deliverable. The prose is what holds up.
Where to start.
Most engagements that fail were sold before the buyer knew what they wanted. Start with a read - the AI readiness assessment guide covers what good looks like, and the free AI Alignment Snapshot is the self-serve version. The longer argument behind both sits in our book, The Elephant in the Algorithm.
Questions people ask.
What does AI transformation consulting actually deliver?
It varies wildly by firm. At the enterprise end (Accenture, Deloitte, McKinsey QuantumBlack, BCG X), it includes business strategy, AI strategy, build, and managed services - typically a multi-million-dollar engagement. At the specialist end (focused change firms, mid-market consultancies), it is a sequenced program of diagnosis, workflow redesign, and adoption work. Ask any firm to define their deliverable in one sentence before contracting.
When is it worth hiring a consulting firm vs doing it in-house?
Hire when the gap is pace, perspective, or political cover - external eyes can name things internal teams will not say out loud. Do it in-house when the gap is execution and the strategy is already clear. The expensive mistake is hiring strategy consultants when you already have a strategy you are not executing.
How much does AI transformation consulting cost?
Enterprise programs from the big firms run from $1M to $25M+ over 12-24 months. Mid-market focused programs from independent firms run from $50k to $500k. The price band that actually fits most $40-150M businesses is the second one - the first is built for organizations 10x larger.
How is AI transformation different from digital transformation?
Digital transformation was about replacing analog systems with digital ones - largely a build-and-rollout problem. AI transformation is continuous: the models change quarterly, the use cases multiply, and value depends on whether people actually change how they work. The change side gets bigger; the build side gets smaller and more iterative.