The task that took your team ten hours now takes four.

Do you cut the fee, keep the difference, or fill the other six hours with more work?

Before you answer, check whether you actually saved six hours. Then check what the client agreed to buy.

Those two questions will get you further than another argument about whether AI has killed the agency model. (For the broader picture of how owners are actually using it, see AI in agencies.)

Measure the finished job

The first draft got faster, which is wonderful. What happened after that?

Count setup, inputs, generation, review, correction, client revisions, and the work required to integrate the output. Include tool costs and any new supervision your process needs.

Compare equivalent work at an acceptable quality level. A rough draft produced quickly and a finished deliverable are different things.

If a strategist spends the afternoon fixing the supposedly automated output, the savings moved departments. They didn't necessarily reach your margin.

Run the comparison over several representative assignments. One unusually easy job is weak evidence for repricing an entire service.

Your agreement changes the answer

If you bill for actual hours, bill for actual hours under the agreement. Don't invoice ten hours because the task used to take ten. If the economics no longer make sense, discuss a different commercial model for future work.

If you charge a fixed project fee, greater efficiency can improve the economics of that project while you meet the agreed scope and quality. It doesn't settle what the next buyer will be willing to pay.

If you charge a retainer, examine what it covers: defined work, ongoing management, access, capacity, or a combination. Faster production affects those arrangements differently.

Start with the promises you've made. Then decide what a better future offer would look like.

Work through the math without pretending it's all profit

Take a hypothetical fixed-fee project priced at $12,000.

Previously, it required 80 hours at an assumed internal direct labor cost of $75 per hour. Direct labor was $6,000, leaving $6,000 before overhead and other costs.

Now suppose the complete job, including review, takes 50 hours at the same cost rate and adds $300 in attributable tool costs. Direct cost becomes $4,050, leaving $7,950 before overhead and other costs.

That's a $1,950 improvement in the project calculation, if the assumptions hold. It isn't automatically $1,950 more cash this month. Salaried labor may still be paid, and freed capacity needs a useful destination.

If you cut the fee to $9,000 under those same assumptions, the amount remaining becomes $4,950. You made the work faster and left less behind than before.

That could still be a deliberate strategy if it creates a better overall business. Just don't mistake a lower price plus faster production for an automatic win.

Choose where the saved capacity goes

You have options, and each should solve a business problem.

Keep the scope and improve delivery economics. Useful when the offer still provides value and your margin needed repair. Make sure quality and service stay intact.

Improve the service. You might shorten turnaround, investigate more thoroughly, or test more alternatives. Be specific about what improves and whether the client values it.

Offer a narrower, lower-cost option. This can make sense for buyers who need less involvement. Define the boundaries so the new offer doesn't become the old service at a discount.

Change the offer entirely. If the original work no longer merits the fee, decide what problem your expertise can now solve profitably. That takes more work than adding "AI-powered" to the same deliverables.

You can also use the capacity to reduce overload. Every improvement doesn't need to become another promise to the client.

Have an answer when the buyer asks about AI

A buyer may reasonably ask why the fee remains the same if part of the work gets easier.

Answer in terms of the actual agreement and what your team is responsible for. Explain the work, review, judgment, or ongoing service the fee covers. Be truthful about the use of AI and follow any disclosure or approval commitments you have made.

Don't hide behind "you pay for value" if you can't explain that value. And don't imply that AI savings are irrelevant simply because discussing them is uncomfortable.

If the buyer can now obtain an acceptable substitute much more cheaply, you have a commercial question to address. Your cost history won't persuade them to ignore it.

Test a pricing decision before rolling it across the roster

Choose one clearly defined service. Establish its current economics and delivery standard. Test the changed workflow, then evaluate the fee and scope against what buyers need and what your team can sustain.

For existing accounts, separate a future offer redesign from changes to an agreed engagement. Have the conversation and document the decision.

The agency owner playbook addresses the wider operating choices. If faster delivery has created a pricing question your team keeps arguing around, bring the numbers to a WTF Call.

You don't need to reward every efficiency gain with a discount or punish it with more deliverables. Decide what kind of agency the saved time is supposed to help you build.

Go deeper: AI pricing is one of a dozen operating decisions landing on your desk. Here is the playbook for the rest. The Agency Owner Playbook