How AI could change A/E fees and billable hours

Why does AI put pressure on hourly billing?#

AI puts pressure on hourly billing because hourly fees pay for time, and AI tools reduce the time some tasks take. If a project team drafts specifications, researches code requirements or produces documentation faster, an hourly contract bills fewer hours for the same deliverable.

The effect shows up first in utilization and chargeability reports. Staff produce the same output with fewer chargeable hours, so either the firm finds more billable work for the same people or revenue per employee falls. A firm that judges performance mainly by billable hours will read a productivity gain as a performance problem.

Clients will notice too. Owners with repeat programs compare invoices across similar projects, and some may ask directly how AI use affects the hours they are billed.

How do hourly, fixed and value-based fees compare under AI productivity?#

Under AI productivity, hourly fees hand the savings to the client, fixed fees let the firm keep them if scope is controlled, and value-based fees separate price from effort entirely. The table compares the models a CFO is most likely to manage.

Most firms will run a mix. Public clients using qualifications-based selection still negotiate fees from estimated effort, so hourly logic persists there even when private work moves toward fixed fees.

Which A/E tasks are most exposed to AI?#

The tasks most exposed to AI are repetitive production and documentation work; the least exposed involve professional judgment, responsible charge and work in the field. Exposure also depends on how much checking AI output requires, because a licensed professional still has to review and stand behind it.

Exposure does not equal savings. Time saved drafting a specification can be lost again if review becomes harder, so measure net hours per task rather than assuming a tool's promised speed.

Exposure also differs by discipline and project type. A firm that produces many similar sheets for repeat building types will see larger effects than one whose work is mostly one-off studies, so compare like with like before drawing firm-wide conclusions.

• 1More exposed: sheet setup and documentation, specification drafting, code research, meeting minutes, first drafts of RFI responses and proposal narratives.
• 2Moderately exposed: quantity takeoffs, clash review triage, energy and load model setup, submittal log management.
• 3Less exposed: client programming conversations, design judgment, coordination decisions, site observation, sealing documents and responsible charge.

What should a CFO measure now?#

A CFO should measure hours by task code before and after AI tools are introduced, using the firm's own timesheet and budget data rather than vendor claims. Deltek Vantagepoint, Deltek Ajera and BQE Core already hold the history needed for a baseline if task codes have been used consistently.

Build the baseline from projects that closed before the rollout, grouped by project type and client, so a change in client mix or a slow season does not masquerade as an AI gain. Then compare new projects of the same type against that baseline phase by phase.

Add one new task code for reviewing AI-assisted output. Without it, review time hides inside production codes and the firm cannot tell whether a tool helps or simply moves effort from drafting to checking.

Why non-fee revenue lines may matter more#

Non-fee revenue lines may matter more because they do not shrink when production hours do. Some firms are looking at productized services, software and tools built from internal expertise, training, advisory retainers and licensing their operating records to AI developers.

Licensing records differs from the others. The firm licenses, rather than sells, records such as RFI histories, review comments and budget vs actual data, and it keeps ownership. Value is unknown until a buyer engages and depends on rights and preparation, so it belongs in planning as a possible additional line, not a replacement for fees.

There is a useful overlap: the records a CFO needs to measure AI's effect on fees, timesheets by task, variance with causes and review histories, are also the records that describe how professional work actually gets done.

Illustrative: an MEP firm tests fixed fees on repeat work#

Illustrative: a fictional MEP engineering firm designs tenant fit-outs for a handful of repeat office clients and bills them hourly. After adopting an AI-assisted specification tool, its CFO sees chargeable hours on those projects fall while output stays the same.

The CFO pulls timesheets by task code from BQE Core and confirms the savings sit in specifications and documentation, not in construction administration. The firm offers fixed fees for design phases on repeat fit-outs and keeps hourly billing for construction administration, where effort still depends on the contractor and site conditions.

The firm also adds a review task code. Within a few projects it learns that one discipline's savings are mostly offset by extra checking, so it holds that discipline's fees steady instead of cutting them.

How SourceX looks at fee and productivity records#

SourceX does not advise on pricing, but the records a CFO assembles to study AI's effect on fees overlap with the records SourceX assesses. Under the SourceX Enterprise Data Value Framework, task-coded timesheets linked to review notes and variance causes can rate well on domain expertise, human-generated signal and AI utility, while preparation cost and privacy burden reduce net value.

If a firm decides to explore licensing, the fit check collects only metadata, and the SourceX five-step transaction keeps the firm's approval in every step, including which financial fields are removed or banded before Delivery.