How Agencies Should Use AI to Win the Next 18 Months

Hero image for Agents of Change featuring founders Gareth Healey (Agency Advisor, right) and Callum Healey (AI Lead, left). Gareth is wearing a yellow polo and glasses, while Callum is in a brown t-shirt. They are seated in grey armchairs, representing a consultancy focused on agency growth and AI adoption.

Most agencies have added AI to their stack. A handful have actually changed their business because of it. The difference between those two groups is not the tools they picked – it is whether they understood what AI makes possible and were prepared to act on it commercially.

This post covers why bolting AI onto existing delivery models is a dead end, what it means to genuinely re-architect delivery around agents, how to reprice services toward strategy and judgement, and why the agencies that do this now will be structurally harder to compete with in 18 months’ time.

Why Bolting AI Onto the Old Model Does Not Work

There is a version of AI adoption happening at most agencies right now. Someone on the team is using ChatGPT to speed up first drafts. A few people have Midjourney or Firefly open in another tab. There might be a Notion AI integration or an automated reporting tool. Everyone is moving a bit faster.

But the commercial model has not changed. The retainer is still priced on hours. The timesheet still runs in the background. The proposal still sells time, not outcomes. And the founder is still the margin pressure valve – working longer to protect the bottom line. This is not transformation. It is efficiency grafted onto a structure that was already under strain. When AI makes your team 30% faster but your pricing model still charges for time, most of that efficiency gain disappears into margin leakage, scope creep, or just cheaper delivery of the same work. You have made the hamster wheel spin faster. You have not rebuilt the wheel. The agencies that come through the next 18 months in a stronger position than they entered will not be the ones that moved fastest. They will be the ones that made a deliberate structural choice – and then repriced around it.

“The question is not which tasks AI can help with. It is which parts of delivery can now be handled by agents – and what that frees your people to do instead.”

What Re-Architecting Delivery Around Agents Actually Means

Re-architecting is not a technology decision. It is an operational and commercial one. It starts by asking a different question. Not “which tasks can AI help us with?” but “which parts of our delivery can now be handled by an agent pipeline, and what does that free our people to do instead?”

For most agencies, a significant proportion of billable hours sit in work that is repeatable, templatable, and context-dependent but not genuinely creative or strategic. Content production at volume. Research and briefing. Reporting. Asset resizing and variant generation. Initial audit work. These are not low-value in the sense that clients do not care about them – they often care a great deal. But they are low-skill relative to what your best people are capable of. When you rebuild delivery around agents handling that layer, two things happen. Your capacity model changes – you can serve more clients or serve existing clients more deeply without proportional headcount growth. And your people are freed to operate at the level where human judgement genuinely matters: strategy, diagnosis, creative direction, client challenge, relationship depth. That is the re-architecture. Not fewer people, necessarily. Different people doing different things – and a delivery model that reflects where the real value is created.

How to Reprice Toward Strategy and Judgement

This is where most agencies stall. They see the operational logic of re-architecting delivery. They might even start doing it. But then they leave the pricing model exactly as it was, because repricing feels like the riskiest move. It is actually the most important one.

If your commercial model still charges for time – even implicitly, through retainers sized by hours – you are capping your margin at the ceiling of human productivity. Agents do not punch a timesheet. When your delivery relies heavily on agent pipelines, a time-based model actively punishes you for efficiency. The shift is toward pricing on outcomes, expertise, and access to judgement. Retainers that reflect the value of the strategic relationship, not the volume of deliverables. Project fees tied to results, not hours logged. Productised services with fixed scopes, clean economics, and repeatable delivery. Premium access to senior thinking, not junior execution. This is not about charging more for the same thing. It is about being clear – with yourself first, then with clients – that what you are selling is no longer time. It is the ability to make better decisions faster, with better outputs as the by-product. That is worth more. It also survives AI disruption in a way that hourly billing simply does not. The agencies I work with that have made this shift are not just more profitable. They are easier to run, less dependent on headcount, and more resilient when client budgets come under pressure. Clients who buy strategy and judgement are harder to replace with a cheaper option. Clients who buy hours are not.

The Competitive Advantage Is Structural, Not Tactical

Here is what this is really about. The agencies making deliberate moves now – on delivery architecture and commercial model – are building a structural advantage that becomes harder to close over time. By the end of 2026, every agency will have AI in the mix somewhere. That is not a differentiator. What will differentiate is whether an agency has rebuilt how it operates and how it prices, or whether it is still running the same model with some AI tools layered on top.

Think about what a properly re-architected agency looks like in 18 months. It has a delivery model that scales without proportional headcount cost. It has pricing that captures the value of expertise rather than the cost of time. Its senior people are doing genuinely senior work, because the repeatable layer is handled. And it has probably already renegotiated a chunk of its client base onto the new model – which means the founder is not explaining the shift mid-relationship, they are already operating in it. That agency is not just more profitable. It is more attractive to clients who want a genuine thinking partner, not a production resource. It is more defensible against downward price pressure. And frankly, it is a better business to run. The Standstill agencies – the ones that added AI to the timesheet model and moved on – will get a short-term efficiency bump. Then the pressure will return, because the structural issues that were always there have not been addressed. They have just been temporarily obscured by slightly faster delivery. STANDOUT agencies use this moment differently. They treat AI not as a productivity tool but as an architectural prompt – a reason to ask harder questions about how they deliver and what they charge, and to act on the answers before the window closes.

The Bottom Line

STANDOUT agencies use this moment differently. They treat AI not as a productivity tool but as an architectural prompt – a reason to ask harder questions about how they deliver and what they charge, and to act on the answers before the window closes.

Frequently Asked Questions

How should agencies use AI to improve profitability?

Agencies improve profitability through AI not by using it to do the same work faster, but by rebuilding delivery around agent pipelines and repricing services toward strategy and judgement. The margin gains come from changing the commercial model – not from efficiency alone.

Should agencies move away from hourly billing?

For agencies integrating AI into delivery, hourly billing becomes increasingly self-defeating – it caps margin at the ceiling of human productivity and penalises efficiency. The direction of travel is toward outcome-based retainers, fixed-fee productised services, and pricing that reflects the value of expertise rather than the cost of time.

What does re-architecting agency delivery around AI agents actually involve?

It means identifying which parts of delivery – research, content production, reporting, asset creation, audit work – can be handled by agent pipelines, and repositioning human effort toward strategy, creative direction, and client judgement. The goal is not headcount reduction but a different allocation of what people do.

How do agencies reprice for AI without losing clients?

The transition works best when agencies reframe what they sell before attempting to reprice. Clients who understand they are buying strategic access and outcomes, not hours, are far more receptive to new commercial structures. Agencies that try to raise prices without changing the value narrative face more resistance.

Ready to move from Standstill to STANDOUT?

Find out how AI consultancy works – and whether it’s the right fit for where you are now.