Agentic AI is making it possible to complete work in hours that once took days. For professional services organizations, that sounds like an obvious advantage.
But consider what happens when a consulting firm completes a 100-hour engagement in 60 hours.
The client receives the deliverable sooner. The team has 40 hours available for something else. And if the engagement is billed by the hour, the firm may have just reduced its revenue by 40%.
That contradiction is becoming one of the most consequential questions facing professional services leaders. As AI agents take on more delivery work, firms must reconsider how they price expertise, measure capacity, develop talent, and capture the value of greater productivity.
During a recent industry discussion, Bo DiMuccio, Distinguished Vice President of Professional Services Research and Advisory at TSIA, and Dylan Connerton, VP of Product at Accelo, explored the implications of AI becoming an active participant in professional services delivery.
Their discussion surfaced a question worth examining more closely: What happens when the time required to deliver a service is no longer a reliable measure of its value?
Listen in to the full conversation, How Agentic AI Is Reshaping Professional Services, now:
Key Takeaways
- AI productivity gains do not automatically translate into revenue growth or higher margins, particularly under time-and-materials agreements.
- Agentic AI creates a new form of delivery capacity that requires different approaches to staffing, pricing, and performance measurement.
- Firms can capture more value by expanding delivery volume, developing new services, and pricing repeatable work around outcomes.
- Human expertise and accountability remain essential, even when agents perform much of the execution.
- The next competitive advantage may come from how effectively firms develop human judgment alongside AI capabilities.
What Happens When AI Makes Your Business More Efficient but Less Profitable?
Under time-and-materials billing, revenue is directly tied to the effort required to deliver work. When AI reduces that effort, billable revenue can decline unless the firm changes how it prices or uses the capacity it has gained.
In a recent TSIA survey of 45 organizations, 25% said their largest AI-related initiative involved using AI to deliver existing services faster.
It's a logical starting point. Firms already understand the services they sell, and AI offers opportunities to reduce the effort required to deliver them. But as DiMuccio pointed out during the webinar, delivering the same work faster can create an uncomfortable financial consequence: fewer billable hours.
The challenge is particularly acute for firms whose pricing and performance models still treat human effort as the primary measure of value.
Consider two firms that use agentic AI to reduce the delivery effort for an engagement from 100 hours to 60.
The first continues billing by the hour. Its revenue falls unless it can fill the newly available capacity with additional work.
The second has established a fixed fee based on the outcome delivered. Assuming quality and scope remain consistent, it retains the agreed revenue while reducing delivery costs.
Both firms achieved the same productivity improvement. Their financial outcomes are very different.
Of course, fixed fees introduce their own risks, including scope uncertainty, inaccurate estimates, and unexpected rework. They are not appropriate for every engagement.
The broader point is that AI is changing what it costs to deliver services, but the way firms charge for that work hasn't necessarily caught up.
Firms that fail to rethink their pricing models may create more value for clients without capturing a corresponding financial return.
What Are Professional Services Firms Really Selling?
The growing disconnect between effort and value forces a more fundamental question about what clients pay for.
"What professional services organizations are actually selling is expertise and accountability." - Dylan Connerton
Clients hire a consulting firm for its judgment, experience, and ability to deliver a reliable result. They expect someone to take responsibility when the work is complicated, the information is incomplete, or the outcome falls short. That responsibility doesn't disappear because an AI agent produced the first draft, analyzed the data, or completed a technical task.
In fact, as execution becomes faster, the expertise required to define, evaluate, and stand behind the work may become more valuable. This creates an opportunity to rethink how engagements are structured.
Many projects contain a combination of predictable and unpredictable work. Data preparation, standard reporting, documentation, and recurring analysis may follow repeatable patterns. Discovery, stakeholder negotiation, complex problem-solving, and decisions involving substantial risk often require more human judgment.
A firm could package well-defined, repeatable components into fixed-price deliverables while retaining more flexible terms for work with significant uncertainty.
That approach preserves a connection between price and client value without pretending that every part of an engagement has the same risk profile.
It also requires firms to understand their delivery processes at a much deeper level: which activities are genuinely repeatable, what expertise each requires, and where exceptions are likely to arise.
How Should Firms Measure Capacity When AI Agents Do the Work?
AI agents can perform work that would otherwise consume employee time, making them a legitimate source of delivery capacity. But you can't measure agent capacity the same way you measure human availability.
Traditional capacity planning is built around people, skills, available hours, and forecasted demand. When agents become part of the delivery process, that model becomes incomplete.
An agent may process multiple tasks concurrently, operate outside normal working hours, or complete a deliverable in minutes. Yet its useful capacity is constrained by other factors: the work it can reliably perform, the context it can access, the quality of its output, and the human oversight required.
Treating an agent as the equivalent of another 40-hour employee would obscure those differences. The more useful question is how much additional work the organization can reliably deliver with its combined human and AI capabilities.
That requires distinguishing between three measures:
- Human capacity: The availability of people with the skills and judgment required to deliver and supervise work.
- Agent-enabled capacity: The additional volume of defined work that AI systems can execute reliably, accounting for limitations, supervision, and rework.
- Realizable capacity: The additional client work the firm can accept and complete after accounting for remaining bottlenecks, demand, and delivery constraints.
These measures are related, but they are not interchangeable.
For example, an agent may reduce the analytical work required for an engagement by 20 hours. If the project still depends on a senior consultant's availability for client workshops and final approval, the project may not finish 20 hours sooner.
The efficiency gain is real, but its effect on overall delivery capacity depends on what happens elsewhere in the workflow.
Connerton made a similar observation during the webinar: shortening project lifecycles can increase the number of engagements a team completes over time, even when individuals cannot realistically manage several complex projects simultaneously.
This distinction matters for resource forecasting. Leaders need to understand where AI increases throughput, where human expertise remains constrained, and whether the organization has sufficient demand to turn newly available capacity into revenue.
The Hidden Cost of Faster Delivery: More Work to Verify
One of the less discussed consequences of AI-assisted delivery is the work required to ensure that outputs are accurate, appropriate, and ready for clients. Producing an answer quickly is not the same as producing a result a professional services firm can confidently stand behind.
Connerton described this challenge through Accelo's own experience with AI-assisted software development. Delivery velocity has increased, but the need for code review and quality assurance has grown as well.
The same dynamic applies beyond engineering.
A consulting team may use AI to generate a financial model, but someone must validate its assumptions. An agency may accelerate campaign development, but its strategists still need to assess whether the work fits the client's market. An IT services firm may automate technical tasks while retaining responsibility for security, reliability, and compliance.
These activities are part of the cost of delivering a trustworthy outcome, and they should be reflected in project estimates, resource plans, and profitability forecasts. Otherwise, firms risk overstating their AI productivity gains by measuring the effort eliminated while overlooking the work introduced.
The relevant financial measure is the net delivery advantage after accounting for AI costs, human review, rework, and the quality of the result.
That is a more demanding standard than counting hours saved. It is also far more useful.
How Is AI Changing the Traditional Professional Services Leverage Model?
Professional services firms have traditionally relied on a mix of senior and junior talent to deliver work efficiently and profitably. Senior experts provide direction and judgment, while less experienced professionals handle research, analysis, and execution as they develop their skills.
AI agents introduce a new dimension to this model, taking on work that previously required human effort.
And that raises a difficult long-term question: If agents perform the work through which junior professionals traditionally gain experience, how will firms develop their next generation of experts?
DiMuccio highlighted this tension during the discussion. Junior employees often develop judgment through repeated exposure to tasks that AI agents can now increasingly perform.
Removing those tasks from the human workload may improve short-term productivity. But eliminating the learning opportunities that accompany them could weaken the firm's future expertise.
The answer may require redesigning early-career roles.
Instead of spending as much time producing initial work, junior professionals may increasingly learn to evaluate AI-generated outputs, identify faulty assumptions, investigate exceptions, and explain why a recommendation is or isn't appropriate.
That still requires structured development.
A junior consultant cannot reliably evaluate complex work without understanding the principles behind it. Organizations will need deliberate approaches to mentorship, progressive responsibility, and exposure to real client decisions.
The implication extends beyond workforce planning.
Human judgment is becoming both a critical delivery capability and one firms must intentionally develop.
An organization that automates execution without protecting its expertise pipeline could eventually undermine the very advantage clients are paying for.
The Professional Services Firms That Will Benefit Most from Agentic AI
The firms that benefit most from agentic AI will likely be those that turn productivity gains into greater delivery capacity, new services, and stronger financial performance.
An important difference exists between employees who know how to use AI and those qualified to deliver AI-powered services to clients.
TSIA's research illustrates just how wide that gap can be. Across the organizations surveyed, the median share of employees proficient in AI was 50%. But the median share qualified to deliver AI as a billable service was just 15%.
That means firms relying on general AI proficiency to assess their capabilities could be overestimating their AI delivery capacity by more than 3x.
And the distinction appears to have real business implications.
When TSIA divided respondents into three groups based on their workforce's AI delivery qualifications, organizations in the lowest group moved approximately 30% of AI pilots into production. The middle group achieved 38%, while the highest group reached nearly 70%.
Although the survey was relatively small, the findings suggest that developing employees who can reliably deliver AI-enabled services may be more valuable than simply increasing AI adoption across the workforce.
For professional services leaders, the question becomes how to build that delivery expertise, and then turn it into additional capacity, new services, and profitable growth.
Three possible uses for the capacity AI creates:
- Increase delivery volume. Complete more engagements with existing teams, provided there is sufficient demand, and the remaining human bottlenecks can be managed.
- Develop new services. Apply expertise to AI implementation, data readiness, evaluation, governance, and other emerging client needs.
- Improve engagement economics. Reduce the cost of delivering defined outcomes while maintaining quality, appropriate pricing, and client trust.
Each offers a different path to financial return. Each also requires decisions beyond adopting AI tools.
A firm cannot assume additional capacity will become revenue without a pipeline to support it. It cannot charge for new services without developing credible delivery capabilities. And it cannot protect margins without understanding the full costs of AI-enabled execution.
The strategic work lies in connecting these decisions.
What Should Professional Services Leaders Reconsider First?
Before expanding agentic AI across their organizations, leaders should test the assumptions behind their current operating model.
Start with a representative engagement and ask:
- Where is the value created? Identify what the client is ultimately paying for and which activities contribute most directly to that outcome.
- Which work can AI reliably perform? Separate repeatable execution from tasks requiring substantial judgment, context, or client interaction.
- What does the full delivery cost look like? Include human effort, AI-related costs, quality assurance, and rework.
- How does the efficiency gain affect revenue? Evaluate whether current contract terms reward or penalize faster execution.
- What will happen to the capacity released? Determine whether it can support additional engagements, new services, or higher-value work.
These questions are useful even for organizations in the early stages of AI adoption, as they connect technology decisions to the economic realities of running a professional services business. They also expose a risk that is easy to miss when AI initiatives are evaluated primarily on productivity: a firm can become dramatically more efficient without becoming more profitable.
RELATED: Test your AI readiness: do you have the operational clarity, data integrity, and decision maturity to gain meaningful value from AI?
What Will Give Professional Services Firms a Competitive Advantage?
The technology is advancing quickly. The business models built around it are proving harder to change.
Professional services leaders now face decisions that will shape how their firms compete: what to charge for work that takes less time, how to account for AI-driven capacity, and where to invest in the human expertise that makes the output valuable.
Those decisions will determine whether AI becomes another source of operational efficiency or a meaningful driver of profitable growth.
Faster delivery is a promising beginning, but the real opportunity is building a business that knows what to do with the capacity it creates.
.svg){kind=small}
.png)
.svg)
