A catchy phrase is making the rounds on social media: the „Consulting Collapse". The claim: the AI wave will not save the big consulting firms, it will expose their structural weakness. That is overstated, but the core is right. It is not a collapse. It is a reshaping, and it is overdue.
For decades one model worked: consulting sold complexity. Long transformation roadmaps, large teams, multi-phase engagements, endless decks. That held as long as companies felt they needed outside guidance to navigate uncertainty. AI changes exactly what clients value.
The numbers tell a clear story
The big firms are restructuring, visibly. McKinsey shrank from around 45,000 people (2022) to about 40,000 (mid-2025), with a further 10 percent cut at the end of 2025. Accenture cut roughly 11,000 roles while committing three billion dollars to AI. Entry-level roles are hit hardest, the classic pyramid of many junior analysts.
At the same time the business model is shifting away from the day rate toward the outcome. McKinsey now earns about a quarter of its revenue from performance-based arrangements. At Bain, AI- and tech-enabled work makes up around 30 percent of the business, trending toward 50. BCG expects a jump from about 20 (2024) to roughly 40 percent AI revenue (2026). An IBM study from 2025 puts it bluntly: 86 percent of buyers want AI-enabled services, and 66 percent would drop providers that do not deliver them.
What clients want today
Companies do not want more decks. They want working systems. They do not want 18-month plans, they want measurable results in weeks. They do not want general advice at premium rates, they want technical depth, speed, and execution. AI devalues labor-heavy busywork and raises the value of small, technically deep teams that actually build, deploy, and improve.
A simple comparison makes it visible. Boutique and AI-native teams often deliver production-ready systems in four to sixteen weeks. Large enterprise projects are built for six to eighteen months and two to four million dollars. And according to Gartner, only 48 percent of AI projects that start ever reach production. Speed and the ability to deliver are no longer nice extras, they are the difference.
When AI works without judgment
How risky it gets to use AI without care was shown by a case in autumn 2025. Deloitte had to partially refund an Australian ministry for a report worth around 290,000 dollars. A generative AI had invented studies that never existed and even fabricated a court quote. It was not the firm that noticed, but an attentive academic.
The lesson is not „AI is dangerous". The lesson is: AI without human judgment destroys exactly what consulting lives on, trust. The value of consulting does not fall. It shifts, away from pushing slides, toward judgment, execution, and accountability for the result. AI-bullish, but not naive.
The big firms now sell AI too
A second strand belongs to an honest assessment. The consulting firms increasingly position themselves as AI distributors. PwC has become OpenAI's first reseller of ChatGPT Enterprise, equipped more than 100,000 of its own staff, and sells access on to its clients. Bain became OpenAI's first partner in the consulting sector. That is legitimate, and the licensing and rollout questions do need support.
But reselling licenses is not the same as delivering value. Rolling out a tool does not yet create a solution. It is exactly into this gap that a new generation of providers is stepping.
The hour of lean consulting
This is the real, almost dramatic shift: it is the hour of small, highly specialized units. Solo professionals, freelancers, lean consulting agencies that deliver with genuine expertise and high speed. AI automates the research, the modeling, the analytical busywork that once required whole armies of juniors. As a result, lean teams can now handle scopes that used to need huge rosters, faster and at a fraction of the cost.
Think it further: consulting, equipped with lateral thinking and high analytical ability, increasingly becomes a software forge itself. Not for interchangeable off-the-shelf software, where buying usually stays right, but for highly individual, sometimes deliberately short-lived solutions that create a real competitive edge. The decision shifts toward make where it creates an edge, and stays at buy where it is commodity.
To stay honest: self-built software also has lifecycle costs, maintenance, security, technical debt. Whoever builds must also be able to operate. That is why what counts is not „build as much as possible yourself", but the clean judgment of where in-house build creates an advantage and where it does not.
What this means for you
The question to consultants is no longer „help us figure it out", but „can you deliver?". For SMEs and public administration that is good news. They now get the reliability once reserved for the big firms, combined with the speed, results focus, and cost advantages of small units. Provided the partner can do both: advise and build.
This is exactly where Vellmerk.ai stands. Vellmerk.ai still advises on the business model, but goes concretely into implementation, in weeks instead of quarters, lean instead of with overhead, and with a curated network of experts, freelancers, and doers tailored precisely to your project.
“The old question was: help us figure it out. The new one is: can you deliver? That is exactly what Vellmerk.ai is built for.” , Thorsten Vellmerk
Sources and further reading
Deloitte partially refunds report with AI errors (Fortune, 2025): fortune.com
AI forces McKinsey, BCG and Bain to rethink fees (TheStreet): thestreet.com
McKinsey layoffs as a warning signal for the industry (Fast Company): fastcompany.com
PwC becomes OpenAI's first ChatGPT Enterprise reseller (TechCrunch, 2024): techcrunch.com
Why enterprises move AI projects from the Big Four to boutiques (Winklix): winklix.com
About Vellmerk.ai
Vellmerk.ai is an AI consultancy (Danish ApS) founded by Thorsten Vellmerk. Drawing on 20+ years of process and IT experience and several years of hands-on AI consulting, Vellmerk.ai helps SMEs and public administration adopt AI in a practical, sovereign way, from strategy to local, on-premise-ready implementation. Proven across multiple client projects. Book an initial consultation.