The Agency Edge: YOUR AI CHALLENGE
Comments Off on The Agency Edge: YOUR AI CHALLENGEFast but flat: Is your agency team becoming over reliant on AI?
“The team’s using AI for everything… but I can’t always see where the thinking is.”
That’s the tension we’re hearing from a lot of agency leaders right now.
Six months ago, the question was, “Should we tell clients we’re using AI?” Now, everyone’s using it. It’s in brainstorms. It’s in research. It’s in first drafts. It’s in the templates your teams are quietly building for themselves. And that’s not a bad thing. In fact, it’s essential. The risk isn’t that AI is in the work but that your people start to remove their thinking from the work.
When speed quietly squeezes out judgement
Under pressure, AI feels like a gift. Short deadline. Tricky brief. Tuesday afternoon brain-fog. You paste in a few bullet points and get a decent first draft. You tidy it, send it and move on.
Individually, that doesn’t feel dangerous. But scaled across a whole team, week after week, something starts to happen. Pitches initially feel polished… but somehow lack a point of view or clients stop saying, “That’s such a you idea.”
The danger here is that alongside saving time your people are quietly outsourcing the thing your clients value the most: Your critical thinking, your judgment and taste!
AI is brilliant at tasks. Your value is in choices.
Used well, AI should absolutely be doing more heavy lifting on repetitive tasks like summarising long documents and identifying themes, turning messy brainstorm outputs into structured options or drafting first version copy or plans.
But only your people can decide which insight actually matters to this client, right now, or say, “This looks good, but it’s not good enough yet” and push a bland idea until it’s sharp, brave and still on-brand.
It’s here where many agency leaders are seeing that gap. How to use AI in a more practical and consistent way without switching off the brain.
From ‘playing with prompts’ to a shared way of working
Most agency teams we work with are already experimenting with prompts. A few people have their own hacks. Some have built little internal libraries.
What’s often missing is a simple, shared way to:
- Brief AI clearly (so you’re not wrestling with vague, generic outputs).
- Break complex tasks into steps, rather than dumping everything in one mega-prompt.
- Apply a quick “Human Check Layer” at the end: Does this actually make sense?
- Does this actually make sense?
- Is it on-brand for us and for this client?
- What’s missing that only we can add?
- Would I stand behind this in a client meeting?
When teams have that kind of framework, the AI speed advantage stays but it also builds-in human critical thinking, so that the agency edge remains.
It’s why we are helping agencies train their people to use consistent AI Prompting methods like COCO. It stands for:
C – Character: Tell the AI the expert it should pretend to be. For example, a B2B Professional copywriter. It helps the AI draw on specific knowledge areas.
O – Objective: What you want to achieve. The AI needs to understand your goal.
C – Context: What else does the AI need to know? All the essential information. Upload examples of what you want and don’t want. Detail matters here!
O – Output: What should it look like? Are there any rules or limits? Example: Each headline must be under 60 characters. Use no buzzwords. Keep the tone urgent but credible.
Now before you push Go, one final instruction:
“Do not guess. Ask me anything else you would find helpful to complete this task better.”
And then answer its questions. It helps minimise hallucinations and gives the AI a chance to ask for any important information that you may have missed.
Credit to Anna Carina Berkman (Architect Anna) for the COCO approach. It’s effective, easy to use and remember!
A quick sense-check for your own agency
Is your agency still in enthusiastic adoption or now in the phase of implementing AI with thoughtful practice?
The agencies getting ahead aren’t banning AI tools or writing long policies. They’re doing two things:
1. Resetting expectations – being explicit that AI is there to amplify their people’s thinking, not replace it.
2. Giving teams practical skills – short, hands-on training so people know how to brief, stack and stress-test AI, without losing their own judgement.
If you’re already feeling that slight drop in sharpness – that sense that work is faster but somehow flatter – it might be time to look at how your people are working with AI.
And if you’d like a practical session to help your team build that “human-first, AI-smart” way of working, get in touch. It’s an area where we’re spending a lot of training time with agencies right now.
