Faster work doesn’t automatically mean a more profitable professional services firm
Tasks that once required significant amounts of manual input can increasingly be accelerated or automated, allowing consulting, accountancy and other advisory firms to deliver work faster and potentially free up valuable professional capacity. Andrew Bone, co-founder and vice president of product , explains why greater productivity from AI doesn’t automatically translate into greater profitability for professional services firms, and what firms should do with their newly available capacity.
AI has crossed an important threshold in consultancy. Almost all firms are now using AI in some capacity. The more interesting question is whether all that investment and activity is actually making firms more profitable.
There is an assumption that productivity and profitability naturally move together. If a task that previously took two hours can be completed in 30 minutes using AI, the productivity gain seems obvious. But consulting firms do not sell individual tasks in isolation. That 90-minute saving might be spread across a week-long project, across several people and dozens of activities. Unless the firm can identify that newly created capacity and put it to productive use elsewhere, the financial benefit can be surprisingly difficult to quantify.
That is the AI productivity paradox consultancy firms now need to solve. To get the most commercial value from AI, firms have to be able to answer a deceptively simple question. What happens to the time AI saves?
If somebody can complete five pieces of work where they previously completed four in the same time, and the firm has the demand and visibility to allocate that additional capacity, there is a clear route to growth. If they simply finish something more quickly but the time saved is absorbed elsewhere and not sold, then the firm as a whole is no more productive.
This is becoming particularly important as the definition of a consulting workforce changes. Firms are increasingly blending permanent employees, offshore teams, contractors and AI within the same delivery model. AI is becoming another resource contributing to the work, but most of the systems and measures firms use to understand capacity were designed for a human workforce.
The costs behave differently too. A person’s salary is relatively predictable. The cost of AI can vary according to usage, tools and workloads, and attributing that expenditure to individual projects is considerably harder. AI costs can fluctuate significantly depending on the models and services being used. It is not always simple to match that spend to a particular client or project, especially with off-the-shelf services. Firms need to understand what AI-assisted work actually costs to deliver if they want to know whether it is improving margins.
This also creates some uncomfortable questions around familiar measures such as billable hours and utilisation. The billable hour is unlikely to disappear overnight. But AI is weakening the assumption that the amount of time something takes is a reliable proxy for its value. If technology enables a firm to deliver high-quality work faster, should the client necessarily pay less for it? You could argue the opposite. There may be circumstances where speed itself carries a premium, particularly if a faster strategy project, transformation programme or piece of advisory work reduces disruption or allows the client to act sooner.
Alternative pricing models do not remove the need to understand time and resources either. In some respects, they make that understanding more important. A fixed fee can quickly become an unprofitable one if a firm does not have an accurate view of what the work costs to deliver.
Human oversight
Against that backdrop, the wider scrutiny of AI-generated research within professional services is important. Recent discussion around major professional services firms has demonstrated the reputational risk when AI-assisted output falls below the standards clients expect. The lesson should not be that firms need to retreat from AI. Speed cannot be separated from quality, judgement and accountability. A report produced in half the time is not more productive if people subsequently have to correct it, or if an error damages a client’s trust in the firm.
Human oversight therefore remains fundamental. Consultancies ultimately sell expertise and confidence. AI can support professional judgement, automate parts of delivery and give people more time to focus on higher-value work, but firms still need appropriate review and governance around the output it produces.
There is also a bigger competitive question emerging. Simply having AI will not differentiate a consultancy for much longer. When competitors have access to broadly similar technology, competitive advantage comes from what firms do with it.
That means moving beyond measures such as the number of AI tools deployed, licences purchased or agents created. Those figures can demonstrate adoption, but they tell us very little about commercial impact. The more useful measures will increasingly be whether AI allows a firm to take on more work, improve margins, deploy its people more effectively, improve the client experience or create services that were previously uneconomical to deliver.
For leaders, and particularly CFOs, that requires much better visibility of the economics of delivery. They need to understand the combination of people and technology being used on client work, the true cost of those resources, where capacity is appearing and how quickly it can be redeployed. Firms have spent years developing ways to measure the profitability of work based on people’s time. They now need to move rapidly towards a broader view of profitability that accurately attributes AI spend too.
AI can undoubtedly make consultancy firms faster. But speed on its own is not a business model. Firms need to connect AI-enabled productivity with capacity, pricing, quality and demand. That is when an hour saved stops being an interesting productivity statistic and starts becoming sustainable commercial value.

