A firm can clear the budget, hire the engineers, and put every tool on the shortlist into production, and still sell in 2026 exactly what it sold in 2023. Ask why the revenue hasn’t moved and the answers sound reasonable: not enough money, not enough people, not enough understanding of the technology.
None of those are the actual problem.
Of the mid-market firms that have already put generative AI into production, 53% say they were only “somewhat prepared” when they did it (RSM Middle Market AI Survey, 2025). These are the firms that cleared the budget and bought the tools. The barriers they name aren’t technical. When CB Insights surveyed enterprise leaders on what’s holding up AI agent deployment, the top blockers were integration with existing systems and internal expertise gaps. Neither blocker is the tools.
Budget and technical depth don’t predict which firms get meaningful earnings impact from AI either.
A firm can buy its way past budget, people, and technology, but not past the decision about what it sells now.
The missing skill is commercial
Ask a partner what AI can do and you’ll get a competent answer. Ask what the firm should sell because of it and you’ll get the same list of tools back.
The typical firm has adopted AI drafting, research, and analysis tools. The technology team understands the tools. The partners understand the clients. Nobody can connect the two: look at a capability and attach a delivery model, a price point, and a client segment to it.
That’s the first blocker: a commercialization problem. It doesn’t get solved by hiring another AI consultant to run a prompt-engineering workshop.
Your team already has a general sense of what AI can do, and the capabilities are expanding faster than anyone can track. The more useful knowledge is how a transition like this plays out commercially, and that knowledge exists: streaming did this to music, cloud did it to enterprise software. What happened there maps onto your services, your clients, and your market.
Default AI adoption breaks hourly billing
If your firm bills by the hour and your team gets faster with AI, your revenue per engagement drops. Among law firms that widely use AI, one in five already report trouble meeting billable targets.
The response in most firms is to keep adopting tools and hope volume makes up for the margin compression. It won’t. Your competitors are adopting the same tools. What looks like an advantage lasts until the next firm installs them, and then the only thing left to cut is the price.
BDO’s managing partner has named the ceiling on what bolting AI onto an existing business model can return. Firms reach it faster than they expect, and after that the model itself is the constraint.
The second blocker is a revenue model that hands the gains straight to the client. No amount of tool investment changes that math. Only changing what you sell, and how you charge for it, does. Across verticals, the firms moving fastest are productizing offerings like subscription comprehensive planning, which uses a pricing model built for AI rather than one stretched around it.
Workflow redesign needs the whole senior team
McKinsey tested 31 variables that predict whether AI produces real business impact, and workflow redesign had the strongest effect of any of them, ahead of tool selection, training investment, and executive sponsorship.
The third blocker is a decision no one person can make. A managing partner can’t redesign the offerings alone over a weekend, and a task force of mid-level people doesn’t have the authority to change the business model.
It takes the people who understand the clients, the delivery, the pricing, and the competitive landscape, working the commercial question in one conversation rather than five. And it takes someone from outside. The senior team brings the domain expertise but rarely the experience of a previous technology transition. Left to its own meetings, the firm takes up AI every quarter and decides nothing, because protecting this year’s billables is the rational choice for every partner.
That conversation ends when the firm can name the new services, the delivery model behind each one, the price, and the client who gets the first call. Anything short of that is a discussion, and a discussion doesn’t change what the firm sells.
Most managing partners would say yes to it. They’ve never sat those people down and worked the commercial question through to a conclusion.
Where this leaves most firms
Budget isn’t the issue. Most firms have spent the money. The people it takes are already on the payroll. The tools exist and work.
What’s missing is the commercial conversation. That’s the work I do with firms, and it isn’t about the technology at all. The question in it is what your firm knows that a client will still pay a premium for once the drafting is free, and what you would have to sell in order to charge for it.
Until someone answers that, the tools keep arriving and the invoice keeps getting smaller. Each quarter looks fine on its own. Then a partner opens the year-end numbers, sees a firm smaller than the one in last year’s, and starts counting back through the renewals that each took a little off the top.