Intelligent Consulting: Why AI-First Agencies Replace Workshops

The model that worked for thirty years

For the last three decades, the dominant model in professional services looked like this. A client had a problem. They hired a firm of senior analysts. The analysts ran workshops, interviewed people, built a current-state picture, built a future-state picture, wrote recommendations, and presented a deck. Then the client either implemented the recommendations with internal teams or hired the same firm to help implement them, at a higher rate.

The pricing reflected the model. Senior partner days were expensive because senior analysis was scarce. Junior analyst days were cheaper but still expensive because the work of going from interview notes to structured analysis took humans a lot of hours. Fifty to eighty percent of a typical engagement's cost was the work of getting from messy inputs to a clean, decision-ready output.

AI collapses that work. Not partially. Not gradually. Almost completely, for the kinds of outputs most engagements produce.

What AI actually replaces in the consulting stack

It helps to be specific about what is collapsing, because "AI is replacing consulting" is lazy. AI is not replacing consulting. It is replacing specific layers of the stack, and the ones that are replaced are the ones the industry spent thirty years building its pricing around.

Typical large-firm engagement hours, six-month transformation scope. The bottom category is where human judgement still lives; the rest is now largely automatable.

The first five categories are ninety percent of engagement hours in a traditional model, and they are all things an AI-first agency now does with a small fraction of the time. Interview synthesis that took a team of three analysts two weeks now takes one operator with the right tools two days. A current-state architecture map that took six weeks now takes three. Document production, which is an enormous cost centre in traditional firms, is a rounding error.

What is not replaced is the last category. Senior judgement. The decision of which insight matters, which recommendation to push, which client relationship to invest in, which trade-off the client is actually facing. That work is still human. It is also approximately eight percent of the engagement by hours, and one hundred percent of the value.

What an AI-first agency actually looks like

We are one of the firms making this bet, so let us be specific about what it changes about how we work.

We run fewer, shorter workshops. A traditional firm might run eight stakeholder workshops in a discovery phase. We run two to three. The rest of the input we gather via voice interviews that are transcribed, synthesised, and cross-referenced automatically. The stakeholder spends twenty-five minutes instead of two hours. The quality of the synthesis is higher, not lower, because we can process every interview the same way, at the same depth.

We write less, we build more. A classic engagement produces a hundred-page deck. Ours produces a working prototype. Not a wireframe. A thing your team can log into and use. The cost of going from concept to working prototype has fallen by roughly 80 percent in our workflow. That changes what the deliverable should be.

We stay. Traditional consulting is built around the phase gate: discovery ends, delivery begins, and those are often two different firms. We are built around the opposite. We engage long, we keep the same people on the engagement from diagnosis through live systems, and we share responsibility for the outcome. The economics of AI-first delivery make this possible because the cost of carrying a dedicated team is much lower than it used to be.

We price for outcome, not for hours. Hours pricing made sense when hours were the scarce resource. In our model, the scarce resource is judgement and the delivery is measured in business outcomes. Most of our client engagements are priced as a fixed fee against a defined outcome, with an ongoing retainer for the partnership after that outcome is hit.

What happens to the big firms

The question clients ask us most often in this conversation is whether the Big Four and the classical strategy houses will adapt. The honest answer is that some will and most will not, and the reason is structural.

A firm with ten thousand consultants has ten thousand careers that are built on the current model. Partners are rewarded on the basis of leverage, which is the ratio of their revenue to the cost of the juniors under them. AI-first delivery breaks the leverage ratio because the juniors are not the cost base anymore. Re-engineering the firm to align compensation with the new model is a five-to-ten-year transition that runs straight into the current partnership's retirement incentives.

Some firms will do it. McKinsey has been the most public about trying. Most will not. The more likely path is that the big firms keep a shrinking discovery-and-strategy practice and a growing implementation practice, and the middle layer of the industry - strategy-heavy mid-sized consultancies - is hollowed out.

The new layer that replaces that middle is small, specialised, AI-first firms. Twenty to fifty people. Deep tool stacks. Long client relationships. Outcome-based pricing. This is the tier the market is building right now, across Europe and in pockets of the US.

What this means for clients

If you are a buyer of professional services, the implication is specific. You should expect to pay less in total for strategy and discovery work than you did in 2022, and you should expect to pay more for implementation and ongoing partnership, because that is where the value now lives.

You should expect any firm pitching a multi-month, multi-million discovery phase in 2026 to have a very good answer to the question "what is an AI-first firm's price for the same scope?" If they do not have one, you are buying an obsolete model.

You should also expect to reorganise your own procurement process. Buying an AI-first engagement does not fit the shape of a traditional RFP. The deliverables are different. The pricing is different. The engagement length is different. Procurement teams that have not updated their templates since 2022 will struggle to contract with the new model, and that will delay your own modernisation.

Our own position

We should be clear about our own stake. We are Calmworks. We are an AI-first agency. We do not claim to be neutral on this. But the data we are citing is not ours. It is the Stanford AI Index, the McKinsey survey, the EIB reports. The shift is visible whether or not we exist to argue for it.

The reason we wrote this piece is that we keep talking to clients who are choosing between "the firm we have used for ten years" and "a smaller, newer thing we do not fully understand." That is a false choice. The smaller, newer thing is not risky because it is new. It is lower-risk because it is priced against outcomes, delivered by people who stay, and built on tools that compress the work the old model billed for.

The risky choice in 2026 is paying strategy-firm hourly rates for work AI can do in days, and then paying a second firm to implement the recommendations the first one will not be accountable for.

If your organisation is about to start a strategy engagement and you want a second opinion on scope and pricing before you sign, we do that in thirty minutes, no engagement required.