Considerations for AI in professional services delivery

Professional services firms have begun integrating artificial intelligence into delivery in a wide variety of ways. The pace of change is significant, and the practices that will eventually become standard are still emerging. Clients commissioning advisory and assurance work are increasingly encountering questions about how AI is used in delivery, what the implications are for the work product, and how to evaluate competing approaches.

This article outlines the considerations that affect the value clients receive from AI-supported professional services and the questions worth asking when evaluating providers. The aim is to support informed decisions rather than to advocate for any particular approach.

What AI can support effectively

Several categories of professional services work are well suited to AI support. Document review and synthesis benefit substantially from current AI capabilities, particularly when the volume of source material is large. Framework mapping and crosswalk work, where deliverables involve aligning content to multiple regulatory or standards frameworks, can be accelerated significantly. First-draft authoring, where the deliverable structure is well established and the content draws from a defined knowledge base, is another area of meaningful capability.

Evidence aggregation across diverse sources, including unstructured documents, system exports, and prior reports, has been a notable area of efficiency improvement. Data analysis and pattern identification across structured datasets can be performed with less manual effort. Translation between technical and business language, including the production of executive summaries from detailed analytical work, is increasingly viable.

What requires human judgment

Several categories of work require human judgment that AI does not currently replace effectively. Materiality determinations, where the question is whether a finding rises to the level requiring escalation or remediation, depend on contextual judgment that incorporates factors specific to the client's situation. Stakeholder management, including the conduct of interviews, the reading of organizational dynamics, and the navigation of internal politics, is fundamentally human work.

Recommendations that affect business decisions, particularly those with material consequences, require professional judgment grounded in experience. Conclusions that go to boards or audit committees carry weight precisely because a qualified professional has signed them. The interpretation of edge cases, where the standard guidance does not clearly apply, draws on accumulated practice in ways AI does not currently replicate. Evaluations of evidence credibility, where the question is not what the evidence shows but how much weight to give it, depend on professional judgment.

The integration question

The value of AI in professional services delivery depends substantially on how it is integrated into the engagement model. Several integration patterns have emerged.

In one pattern, AI tools are used by individual practitioners as productivity aids without changes to the engagement model, scoping, or deliverables. The benefit accrues primarily to the firm in the form of efficiency. The client receives a comparable engagement at comparable cost.

In another pattern, AI integration changes what is included in the engagement scope. The same engagement fee covers more depth, broader coverage, more current evidence, or extended ongoing support. The benefit accrues to the client in the form of expanded value at consistent cost.

In a third pattern, AI integration enables engagement models that were previously economically infeasible. Continuous assessment maintenance, ongoing controls monitoring, and other recurring engagement structures become viable when the marginal cost of refresh is reduced. The benefit accrues in the form of new service offerings.

These patterns are not mutually exclusive. The relevant question for clients is which patterns the provider is operating under and what that implies for the value the client receives.

Questions worth asking providers

Several questions help clarify how AI is integrated into a provider's delivery and what the implications are for the engagement.

What does the provider include in scope that would not be feasible without AI support? A provider that has restructured engagements to take advantage of AI capabilities can typically answer this clearly. A provider that has not made such adjustments may answer in general terms.

Who reviews AI-generated content before it reaches the client? Effective integration includes named individuals with the relevant qualifications reviewing specific categories of output. Less effective integration delegates review to junior personnel or relies on process compliance rather than substantive review.

What categories of work does the provider explicitly not delegate to AI? A provider with a clear position on the boundaries of AI use has typically thought carefully about quality. A provider that frames AI as a general-purpose enhancement may not have established such boundaries.

How are clients informed about AI use? Disclosure in engagement letters, statements of work, or other engagement documentation supports informed client decisions. Lack of disclosure raises questions about whether the firm is comfortable with how it is using AI.

What client data is provided to AI services and under what terms? AI integration often involves passing client data through AI services. The terms under which that data is handled, including training and retention provisions, warrant explicit attention.

The pricing question

Clients sometimes raise the question of whether AI integration should be reflected in pricing. The answer depends on which integration pattern the provider is operating under.

Where AI is being used as a private productivity aid without changes to scope or deliverables, the question of pricing is fair to raise. The provider's cost structure has changed, and clients may reasonably expect to see some of that benefit.

Where AI integration has expanded engagement scope or enabled new service offerings, the pricing reflects expanded value rather than reduced cost. The relevant comparison in this case is between the value received and the value that would have been received under a non-AI-integrated alternative, not between current price and a hypothetical lower price.

Distinguishing between these cases requires the kind of clarity from the provider that the questions above support.

The continuing role of professional standards

Professional standards continue to apply regardless of how AI is integrated into delivery. Audit standards, advisory standards, professional ethics, and regulatory expectations are not modified by the use of AI. Providers remain responsible for the quality of their work, the appropriateness of their conclusions, and the protection of client information. AI integration is a means of delivering against these standards, not a substitute for them.