AI hasn't replaced our agency. It's made us a better one
Last month we wrote about why the explosion of AI-generated content doesn't make communication less important. If anything, it makes clarity, trust and human judgement even more valuable.
Since then, we've continued experimenting with AI at Cosy Meerkat, weaving it deeper into the way we research, analyse and develop work for clients. And it's led me to another conclusion that I wasn't expecting.
AI hasn't replaced our agency, and I don’t think it ever will. What it has done, though, is make us a better one.
Depending on who you listen to, artificial intelligence is either the answer to everything or the beginning of the end. Every week seems to bring another prediction about which jobs will disappear, which industries will be transformed, and which organisations are moving too fast or too slowly. Agencies, apparently, are somewhere near the top of the endangered species list.
Perhaps they are. But I suspect many of those conversations are focused on the wrong thing.
At Cosy Meerkat, we've embraced AI wholeheartedly, using tools like ChatGPT and Claude extensively every day. If you looked at how we work today compared with two years ago, you'd see AI appearing throughout much of what we do.
What you probably wouldn't see is AI replacing the things our clients actually value us for.
In fact, the more we use AI, the more convinced I become that it has exposed something actually quite interesting. Much of the conversation assumes that the value of agencies, consultants and communicators sits in the outputs we produce. The report. The strategy. The communication plan. The workshop deck. The content.
Those things are important, of course. They are visible. They are tangible. They are what clients receive at the end of a project. But I've never really believed that's where the value of what we do actually lives.
Over the years, some of the most important moments in projects I've worked on have happened long before a single slide was created or a recommendation was written down. They've happened when somebody has asked an awkward question that nobody else wanted to ask. They've happened when a client has casually mentioned something in conversation that revealed the real problem sitting underneath the one everyone thought they were solving. They've happened when a workshop has taken an unexpected turn because people finally felt comfortable enough to say what they were actually thinking. And they’ve happened when someone said a throw away comment in a meeting which was actually strategic gold.
None of those moments appear in the final deliverable. Yet they're often the reason the deliverable works.
One of the first things we noticed when we started using AI properly was that it was incredibly good at helping us move beyond the blank page. It could help us explore ideas more quickly, analyse information faster and challenge our assumptions in ways that were genuinely useful. However, the quality of what came back was heavily influenced by the quality of what went in.
The better we understood the challenge, the better the results. The more context we provided, the more useful the outputs became. The more experience we brought to the conversation, the more valuable AI became as a tool. Far from replacing expertise, it seemed to reward it. Last month we argued that AI was making human communication more valuable, not less. I'm beginning to think that's only half the story. It's also exposing where that value really comes from.
That's one of the reasons we've spent time creating client-specific Gems and building AI into our briefing process. Every organisation has its own language, history, culture and complexities. The more context we can bring into the conversation, the more useful the technology becomes. So we add tone of voice documents, previous work we have done, as much rich information as we can, before we ask anything of a tool. I should also add that our approach is grounded in strict ethical guardrails. We handle all client data with total privacy, and though AI is an incredible collaborator for us, human judgement always has the final say.
And perhaps that's where my thinking has shifted most over the last couple of years.
What fascinates me about AI isn't really the technology anymore. The technology is impressive, but increasingly I find myself drawn back to the same questions I've spent most of my career exploring.
Why do people embrace some ideas and resist others?
Why do some leaders create trust while others struggle to build it?
Why do some change programmes gain momentum while others stall?
Why can two people hear exactly the same message and walk away with completely different interpretations?
Those questions haven't gone away. If anything, AI has made them more important.
I've sat in enough workshops, leadership meetings and change programmes over the years to know that organisations are rarely held back by a lack of information. Most people already have more information than they know what to do with. What they often lack is clarity. They lack confidence. They lack a sense of how all the pieces fit together.
That's where communication comes in. And I don’t mean communication in the sense of newsletters, intranets and leadership announcements, but communication in its broader sense. Helping people make sense of what is happening around them. Helping them understand where they fit. Helping them navigate uncertainty.
AI can support that work, but what it can't do is build trust. It can't walk into a room and sense that something isn't landing or spot the hesitation in a conversation or recognise when people are nodding politely while thinking something completely different.
Those things remain stubbornly human.
Which brings me back to the original claim that AI is coming for agencies. Maybe it is coming for agencies whose value sits primarily in producing outputs. But I don't think that's where the future lies.
The more I look at the work Cosy Meerkat does, the more I think our role is becoming more and more about creating understanding. Less about producing deliverables and more about helping organisations make sense of complexity. Less about information and more about insight.
Ironically, AI may be accelerating that shift.
By taking away some of the friction, it allows us to spend more time doing the things that humans are uniquely good at: listening, challenging, connecting ideas, building trust and helping people navigate change. Perhaps that's the real opportunity.
We've spent the last two years asking what AI can do. Maybe the more interesting question is what it allows us to do. The more we use it at Cosy Meerkat, the more convinced I become that our future won't be defined by artificial intelligence alone. It will be defined by what happens when artificial intelligence meets human judgement, empathy, curiosity and creativity.
