Two years ago, closing a session on the digital transformation of consulting, I left my students with six open questions — I didn’t have the answers, and neither did the industry. This post is the scoreboard I couldn’t fill in then.
2024 open questions on consulting
It was a little more than one year after the ChatGPT release, and I wrote a first blog post about How Generative AI is Reshaping Consulting. I described three main opportunities for consulting firms: selling AI services to clients, using the technology in their operations, and developing new business models. I also surfaced some dimensions of strategic adaptation, whether forging partnerships or redesigning the organisation.
After that, I developed a more detailed framework for listing the open questions about consulting triggered by Generative AI. I used Bonnet and Westerman’s digital transformation framework and applied it to consulting. I came back with 6 open questions I used in many courses with my students to discuss consulting in particular and the digital transformation of an industry at large:
- Will the dominant business model of time-based billing remain?
- How will the value proposition of consulting change?
- Will new forms of customer experiences emerge?
- How will the value be split between clients, consulting firms and tech providers?
- Which make/buy/partner strategies will be implemented?
- How will the human resources processes change?
If I were to sum up with one single root cause of all these questions, the major impact of AI in consulting is that it commoditises analysis. Everything else follows: the pricing strategy, the value position, the organisation, the competitive dynamics, and the internal processes. It’s one single phenomenon with 6 faces.
Where do we stand 2 years later?
The billing unit is breaking.
Though most revenue is still time-based, we’ve seen global players collect on different units. It’s reported that 25% of McKinsey’s global fees are now outcome-based rather than hourly. PwC launched PwC One, an AI platform that lets clients access six automated services sold on a subscription model (tax analysis, financial reporting, sustainability assurance, deal diligence, and operating model transformation), “without a PwC person in the loop”. Neither alternative revenue model was born with AI, particularly the outcome-based one, which flourished in the early 2000s as operations consulting grew. Similarly, selling access to proprietary software has been one pillar of McKinsey’s business model for decades. AI is accelerating and democratising an existing trend. However, what’s new is that the pricing change is dragging the capital structure with it. McKinsey is shifting partner pay from cash toward equity, because outcome-linked fees are lumpy and deferred and the firm now needs capital inside to carry the risk and fund the technology. An AI-driven pricing change has become a capital-structure change.
The value proposition retreats to what doesn’t commoditise
Consulting has never been about analysis only. Clients buy consulting services for judgement (which option is right for this firm, this moment), accountability (an external name on the recommendation, so the executive isn’t the only one carrying it), and change management (additional brain and muscle for implementing recommendations and transformation projects). All of this is supported partly by some analysis. If analysis is the layer AI zeroes in on, its contribution to the value proposition reduces. Still, the other facets of the consulting value proposition remain and, in some respects, are even more valuable. “AI can write the deck; it cannot take the blame.”
The customer experience didn’t change much.
Except for subscription-based intelligence offers such as PwC One, the customer experience hasn’t changed much so far. On one side, productised self-service (PwC One) is a genuinely new, lower-touch, democratised experience; the client consumes consulting without a consultant in the room. On the other, if what clients ultimately buy is accountability, human judgment and change management, the experience becomes more relational, not less. Decks, presentations and workshops are far from fading away.
Significant value is shifting to the two ends of the chain — tech providers and clients — squeezing the consulting middle.
On the value-capture side, we see pressure on consulting firms on two fronts: pricing and cost/competition. On pricing, more and more clients now ask consulting firms to quantify AI’s contribution and price the return on investment, with medium-term referencing grids distinguishing “replacing” vs “augmenting” AI at different rates. This might push the prices down. Attracted by the juicy consulting market and eager to capture the maximum value of their technology, model providers have launched consulting arms to help clients using their technology as a typical consultant would. As of early May 2026, Anthropic, Blackstone, Goldman Sachs, and Hellman & Friedman have announced a new $1.5 billion partnership to create an AI-native enterprise services firm, aiming to compete directly with traditional consulting firms. OpenAI launched a subsidiary with the same aim that same week.
Partnering has been the reflex, because it’s the fast route to AI resources — but it’s fragile when the partner is also the rival.
In the first years of GenAI adoption, the default mode has been partnerships. All big consulting firms forged alliances with model providers to help their clients implement the technology. OpenAI even formalised these partnerships through “Frontier Alliances” with BCG, McKinsey, Accenture, and Capgemini. This helped both sides to grow the market rapidly. Then, when it became attractive enough, model providers chose to “build a competitor” over pure supply. This is a classic supplier-turned-rival situation in the battle for value capture on the value chain. We saw that movie before in other industries, for example in retail with direct-to-consumer brands going around retailers and selling directly to consumers. The question is whether the producers (here, the model providers) can provide the other dimensions of the value proposition (judgement, accountability, and change management) faster than the distributors (consulting firms here) can develop efficient models internally.
Processes are being agentified, and the org chart is slowly flattening from pyramid to diamond.
McKinsey now runs roughly 25,000 AI agents alongside roughly 40,000 employees, up from about 3,000 agents two years ago. The original target of one agent per employee by 2030 now looks like a matter of months rather than years. In addition to analysis tasks, McKinsey’s agents now staff consultants onto engagements with global rollout targeted for end-August 2026. With lower demand for lower-level analysis tasks, the traditional pyramid model of consulting is under pressure: fewer juniors are needed to do the grunt analytical work. However, even if several big consulting firms reportedly cut their graduate intake, it’s still unclear if AI is the only reason for this cut. In addition, the question of how to grow expertise remains: how will future experts be trained if fewer consultants join consulting and no longer develop their analytical acumen through analysis work? This is why we also see reports of consulting firms maintaining their junior recruitments and developing new training processes. EY launched a new paid “Career Residency” for its assurance interns that will continue for 8 to 12 months after they complete the traditional eight-week internship. EY is launching the residency as AI reshapes the skills entry-level workers need and pushes firms to rethink how quickly new hires can contribute.
Conclusion
AI-driven commoditisation of analysis has several consequences for the consulting industry and business models. The billing unit breaks, and alternatives grow. The value proposition retreats to judgment, accountability and change management. Significant value flows to both ends: tech and clients. Partnerships are inevitable but become a trap when the partner is the rival. Internal processes are automated, and the pyramid flattens.
These trends are real, but they don’t mean the situation is or will be the same for every company in the industry. The market has always been somewhat fragmented with specialised players and more global ones positioning very differently. This will not change as AI grows; quite the contrary. We’ll likely see different models coexisting in the industry as consulting companies will use AI differently in their operations and value propositions.
