Having worked in agencies for the majority of my career,  one thing is clear to me: the traditional client-agency model wasn't built for the world we operate in now. 

With the tsunami that is AI and all that this remarkable technology is bringing, the time has come for a new paradigm. And while I don’t think we ought to throw the proverbial baby out with the bathwater, I do think agencies have to fundamentally restructure client relationships and determine what those relationships should look like.

One thing I want to make clear. Lately, I’ve been hearing this narrative that says AI will make agencies obsolete. That clients will just “prompt” their way to a finished product, cutting out the middleman entirely. The reality is that's not what's happening. It might interest some AI platforms to flog this story, but what is taking place is far more complex and nuanced. 

Yes, AI is upending the old paradigm in which a client requests that an agency with expertise in, say, content creation, draw up a plan and subsequently execute it, showcasing its knowledge and expertise to the client's delight (hopefully) and satisfaction. Typically, this kind of engagement could span months, even years, and so relationships developed and agencies were secure in the knowledge of a guaranteed revenue for an extended period of time. 

AI is obliterating this traditional time and cost of this model. Tasks that once took days - certain types of research, data analysis, design iterations, coding - now take hours. And that collapsing of timeframes is only going to increase as AI evolves and agencies that pretend otherwise are in denial.

However, while this acceleration means shorter project timelines, the need for expertise remains. As someone said recently, while I can use ChatGPT to tell me how to fix my car, I’m not a mechanic. There is still a need for people with a deep understanding and experience of a particular subject or skill. 

And this is where the nuances live: the human, or humans, in the client-agency relationship. The people who understand a client's ambitions, their audience's psychology, or the specific tension between where they are and where they need to be. People who will challenge a brief that's solving the wrong problem. People who can sit in a room with a stakeholder and know, from the way they hesitate before answering, that the real blocker is internal politics, not budget.

Thinking this is only about reduced timelines is overly simplistic. The agencies that are leading the way aren't just using AI to work faster. They're using it to work differently, and to offer clients a fundamentally different kind of relationship.

One that is continuous and multifaceted: a strategic partner, focused on shared outcomes, offers continuous intelligence, and is a capability builder. 

Your agency is now your strategic partner

Now that tasks happen faster and, as a consequence, more cheaply, an agency's value is no longer in execution alone. The real differentiator becomes strategic clarity: the ability to define the right problem, design the right solution, and steer the work when it drifts. That's a partnership, not a supplier relationship.

For starters, the sheer volume of AI tools now available across design, engineering, content creation, audience intelligence and performance optimisation is huge. Deciding what to use and how best to use it within your organisation requires continuous oversight and operational management to deliver value. This is not a one-and-done situation; AI cannot be switched on for a campaign and switched off again.

Yes, when built and deployed properly, it is worth considerably more than any single campaign it enables. But, and it’s a big but, it requires maintenance, iteration, and expertise to operate. 

Delivering it as a project and walking away is akin to building a Formula 1 car and leaving it without a pit crew.

What does this look like in practice?

The best version of this model isn't just strategy conversations and quarterly check-ins. It's an embedded operation that innovates new products and concepts, builds and rebuilds digital experiences, makes your existing assets AI-ready and discoverable, automates testing and quality control, and tracks real outcomes through live dashboards, not vanity metrics in a slide deck. 

I like to think of it as having a specialist digital team inside your business, running continuously.

Your agency should be focused on shared outcomes

If AI compresses delivery timelines, pricing models built on time-and-materials start to look like a rounding error. The new model prices are based on the impact on the results delivered, not the hours logged. 

That's a harder conversation, but it's the right one. The practical implication of this is a new model built around a monthly subscription rather than a project fee. It's the flexibility of a technology subscription applied to design, engineering and optimisation expertise, and it's a significant departure from the way agencies have traditionally structured engagements.

The continuous model is not necessarily more expensive than the project model, but the maths is different, and it requires a different kind of budget planning. Rather than spikes of spend around product and experience launches, you are investing in a consistent monthly commitment that funds infrastructure, intelligence, and creative capacity that compounds over time.

It’s not just about the money either. It’s about responsiveness. Moving away from static deliverables toward dynamic, continuous delivery means brands can keep pace and even get ahead of market velocity. AI-powered analytics and dashboards mean there's no longer any excuse for a client to wait until the quarterly review to find out what's working, or not, as the case may be. Expectations have changed, and clients are looking for more regular input and insight, and importantly, recommendations for change and course correction.

Your agency is a capability builder 

The best agencies aren't hoarding knowledge; they're actively building their clients' internal capabilities alongside their own. An agency that helps you understand and use AI is worth more than one that mystifies it. For example, one area generating significant attention right now is AI-powered engineering. Our CTO wrote about this, and what the impact these new technologies are having on the roadmap in AI-native engineering discipline are radically reshaping the build process and roadmap velocity. In a nutshell, we are seeing many organisations that built their digital estate five or ten years ago finding it increasingly incompatible with where they need to go, but having neither the time nor the budget for massive rebuilds. However, it’s no longer the only option and with AI, the ability to rewrite legacy systems, modernise architecture, and rebuild digital infrastructure is faster than ever.

And if your agency is not talking about this and offering these services, then it’s time to rethink who your partners are.

The organisations that will win, clients and agencies alike, are those that understand where AI ends, and human expertise begins. AI can generate options at scale. Humans decide which option is right. AI can surface patterns in data. Humans interpret what those patterns mean for a real business with real constraints. AI can produce. Humans create meaning.

We recently worked with one of our clients, Fáilte Ireland, to introduce a more personalised search experience when on visitdublin.  AI helped us to deliver, but the responses and the experience of finding inspiration and information for a trip to Dublin were all very much human-led. 

And the results were impressive. The AI Planner generated 250% more usage than the site's standard search function while meeting the standards expected of a public body deploying AI in a transparent and responsible way. 

Read the full case study here https://allhuman.com/work/failte-ireland-ai 

There's another dimension to the human layer that doesn't get talked about enough: governance. As AI becomes embedded in how organisations operate, the question of who is accountable for what it produces matters enormously, particularly in regulated sectors or public-facing organisations. The EU AI Act is already setting expectations, and that's only going to increase. The human oversight in a well-structured AI-augmented model isn't just about creative judgement. It's about ensuring the technology is validated, compliant, and continuously overseen by people who are accountable for the outcomes it delivers.

Is it time to have a talk with your agency?  Probably.

Here are my suggestions for what to ask them.

• How is AI changing your process, and what does that mean for what you charge?
• How are you measuring your work against outcomes, not just outputs?
• Are you investing in understanding our business, or just delivering to a brief?
• How are you helping us to become more capable, or more dependent?
• What AI infrastructure will you build for our brand, and what happens to it if we end the relationship?
• How do you measure success on a rolling 90-day basis, not just at project completion?
• What does your team look like in month six versus month one? Do we get more embedded expertise or more junior resources over time?
• How do you operationalise rapid response? What does it take for you to produce and publish something within 24 hours?
• How does our brand knowledge transfer back to us in a usable format if we ever move on?

The client-agency relationship isn't broken beyond repair. But it does need to be rebuilt on honesty, shared accountability, and a clear-eyed view of what technology can and can't do.

The continuous partnership model is not simply "a retainer" in the way the industry has historically understood retainers, i.e. a vague monthly fee that funds an account team to be available. The best version of this model is structured, accountable, and explicitly tied to ongoing strategic and executional output.

Think of it less like hiring an agency, and more like embedding a specialist team, one that is always on, always learning, and always operating your brand's marketing infrastructure as a living system rather than a series of completed projects.

The agencies worth working with know this. They're already doing it.

The question is whether you're ready to demand it.