the:big:picture
Will AI replace agencies? No, but it will reprice them.
By Loan Laux · July 8, 2026 · 6 min read
AI will not replace agencies, but it is repricing them. The hours agencies billed for production (first drafts, boilerplate code, routine assembly) are collapsing in cost, while the things clients actually buy (judgment, accountability, coordination, taste) are getting scarcer and more valuable. Agencies that keep selling hours will shrink. Agencies that sell outcomes won't.
That's the thesis. Here's the reasoning, what history suggests, which agencies are genuinely at risk, and what to do about it in the next 12 months. Written by someone who built, sold, and operated agencies, not by a futurist.
The honest version of the answer.
Some agency work is genuinely going away. If a deliverable is a commodity (a first-draft blog post, a template site build, resize-and-reformat production), its price is heading toward the cost of a model call plus review, and no positioning deck stops that.
But agencies were never really selling those hours. They were selling the surrounding structure: someone who understood the goal, chose the approach, coordinated the specialists, caught the problems, and answered for the result. AI compresses the middle of that sandwich. It doesn't supply the bread.
What clients buy that AI doesn't sell.
- Accountability. A signature under the outcome, and a phone that rings when the launch breaks at 6pm. Clients pay heavily for someone else to hold the risk.
- Judgment before the prompt. Knowing what to build, for whom, and why: the decisions upstream of any tool. Wrong strategy executed brilliantly by AI is still wrong.
- Coordination. Multi-discipline delivery across design, dev, content, and media, with tradeoffs made coherently. This is management, and it's stubbornly human.
- Taste. The eye that says 'close, but not yet'. Models raise the floor of production quality; the ceiling still belongs to people with a point of view.
The billing-model problem.
Here's the mechanism that actually hurts agencies: hourly billing converts your efficiency gains into your client's discount. If AI cuts a 40-hour deliverable to 12 and you bill hours, you just cut your own invoice by 70 percent.
The agencies handling this well are moving revenue to structures that price the outcome rather than the effort: fixed-scope projects (where efficiency becomes margin instead of a discount), retainers priced on the value of coverage, and productized services with a defined deliverable and price. None of this advice is new. AI just made it urgent.
What history suggests.
Agencies have been declared dead before: website builders were going to end web shops, stock libraries were going to end commissioned photography, programmatic was going to end media agencies. Each wave killed some commodity work, pushed agencies up the stack, and expanded the total surface of work that needed coordinating. Note what actually died each time: agencies whose entire value was the thing being commoditized.
The AI wave is bigger and faster, and the honest caveat is that nobody knows its ceiling. But the structural pattern (commodity production dies, orchestration migrates up) has held through every previous platform shift, and current client behavior matches it: budgets shifting from production line items toward strategy, integration, and AI implementation itself.
Which agencies are actually at risk?
The risk isn't evenly distributed. High risk: shops selling undifferentiated production at rates already under offshore pressure, content mills, template implementers, dev shops whose pitch is a rate card. Low risk: agencies with owned relationships, an industry specialty, accountability for outcomes, and a service mix that already includes advice.
The uncomfortable question isn't 'will AI replace agencies'. It's 'is what my agency sells a commodity with an agency wrapped around it'. If the answer is yes, AI isn't the cause of the problem. It's the deadline.
What to do in the next 12 months.
01:
Adopt internally first
Margin is the cheapest place to win. Map your delivery, change two workflows a quarter, and bank the hours. Our adoption guide for small agencies is the long version.
02:
Fix the billing model
Move new work toward fixed scope and retainers. Reprice the deliverables whose cost structure has visibly changed before a competitor does it for you.
03:
Repackage what clients actually value
Put strategy, coordination, and accountability on the invoice as the product, with production as an input. Proposals should read like outcomes, not hour manifests.
04:
Add an AI services line
Client demand is already at your door. Answer it deliberately, in-house or white-labeled, rather than declining it into a competitor's pipeline.
The realistic picture a few years out.
Extrapolating conservatively from the agencies we work with: teams get smaller per dollar of revenue and more senior per head. Production layers thin. Review, direction, and client-facing layers don't. Services sell as outcomes with AI embedded in delivery, and the agencies planning headcount and pricing for that shape now will experience the shift as an upgrade rather than a crisis.
If you want a concrete read on where your agency stands, that's what an audit produces: a map of your delivery, and the order in which to change it.
Frequently asked questions.
Will clients take AI work in-house instead of hiring agencies?
Some production, yes, the same way some took web updates in-house once CMSes matured. What stays external: work needing outside perspective, multi-discipline coordination, surge capacity, and accountability. In-house AI also needs adoption help, which is itself agency work.
Should agencies lower prices as AI reduces effort?
Reprice deliverables where the market already has (commodity production), and hold or raise prices where you sell outcomes and accountability. A blanket discount gives away margin you need for the transition. Reprice line by line, not across the board.
Which agency roles change most?
Junior production roles change most, and that creates a real problem: the entry-level work AI absorbs is how juniors used to learn. Agencies have to rebuild the apprenticeship deliberately, through review work, client exposure, and AI-supervised production. PM, strategy, and senior craft roles gain ground.
Is it too late to start adopting AI?
No. Most small agencies are still early: scattered individual usage, no changed workflows. Systematic adoption over one quarter puts you ahead of the median agency you pitch against. The window that's closing is the one where adoption is a differentiator rather than the baseline.