The quiet chaos inside your team's AI habits

I spend a lot of time inside creative teams. Agencies, and in-house marketing departments. And when I ask them how they're using AI, I get a version of the same answer, almost every time.

"Everyone's using something. But we're not really sure what."

And there it is. The quiet chaos.

It doesn't announce itself loudly. There's no single moment of crisis. It's more of a creeping feeling—the sense that your team is moving fast but not quite in the same direction. That brilliant work is happening in pockets, but it isn't sticking. That AI is in the room, but nobody's really in charge of it.

If any of that resonates, breathe. You're not behind. You're not broken. You're just very, very normal.

McKinsey estimates that organisations are, on average, only at 1% maturity in their AI adoption. Which means the vast majority of teams, including your sharpest competitor, are still figuring this out. 

There is time to do this right.

So let's talk about what "not doing it right" actually looks like. Because in my experience, it usually comes down to three things.

Three problems I see every week

1. Nobody knows who's using what, and the good work keeps disappearing

When AI tools are chosen by personal preference rather than team policy, something quietly damaging happens. Your copywriter loves one LLM. Your strategist swears by another. Your designer has a third running in a browser tab. And on the surface, that looks like a team that's embracing innovation.

But dig a little deeper and the cracks appear.

The brilliant prompt your copywriter spent three hours refining? It lives in their personal account. It won't be there when they go on holiday. It won't be passed to the next person who takes on that client. It's institutional knowledge, built on your company’s  time, that you will never actually own.

And then there's the IP risk. When you're feeding client briefs, brand guidelines, and strategic data into tools that haven't been vetted, you're potentially walking that information outside the boundaries of your business. Most teams haven't thought about this yet. That's not a criticism, it's just where we are. But it's a conversation worth having before it becomes a problem.

Great prompts are an asset. Treat them like one.

The fix is simpler than it sounds. You don't need to standardise on a single tool overnight. But you do need a shared home for the tools your team agrees to use, and a system for capturing, naming, and storing the prompts that actually work. Think of it as a prompt library. Your team's collective intelligence, organised and accessible. Hours saved. Knowledge protected.

2. There are no rules (and that's scarier than it sounds)

Most creative teams running AI today have no user policy in place. No guidance on what data can go in, what tools have been approved, or what the output review process looks like. It's the Wild West, dressed up as efficiency.

I'm not here to make this feel heavy. Policies don't have to be long. They don't have to be written by lawyers. But they do have to exist.

Because without one, every person on your team is making their own judgment calls. About what's safe. About what's acceptable. About what counts as "their own" work. And that inconsistency, more than any single mistake, is what erodes trust. With clients. With leadership. With each other.

A good AI usage policy answers three questions:

• What tools are approved? Give people a clear, short list of tools the business has vetted. Not to restrict creativity, to protect it.
  
  
• What data is off-limits? Client confidential material, personal data, unreleased campaign strategy. Write it down. Say it out loud. It removes the ambiguity that causes problems.
  
  
• What does good output review look like? AI doesn't fact-check itself. It doesn't know your brand. Build a human review step into the process, and make it non-negotiable.

One page. Clear language. Reviewed together as a team. That's it. And the ROI? You sleep better. Your clients trust you more. Your team feels like grown-ups, not guessers.

3. The scattergun approach—trying everything, mastering nothing

This one is the most common. And honestly, the most human. There are hundreds of AI tools being released every week. They're exciting. They're shiny. And the pressure to keep up, from clients, from leadership, from LinkedIn, is real.

So teams try everything. A little bit of this, a little bit of that. And what they end up with is a patchwork of half-learned tools, inconsistent outputs, and a faint but persistent feeling that they're doing it wrong.

They're not doing it wrong. They're just doing too much at once.

The teams I see making real, compounding progress with AI are the ones who have chosen deliberately. Not the most features. Not the most hyped. The tools that solve their specific pain points, used well, consistently. Depth over breadth. Every time.

So ask yourself: what is the one task that drains the most time from your team each week? Research? First drafts? Briefing documents? Start there. Build a workflow around that one use case. Make it excellent. Then—and only then—move to the next.

That's not caution for its own sake. That's how you actually build something that lasts.

So what does doing it right actually look like?

Diagnosing the chaos is the first step. But knowing what's broken doesn't automatically tell you how to fix it. KPMG's 2024 Technology Survey found that 52% of AI projects stall in pilot or fail to scale entirely — not because the tools don't work, but because teams jumped to complexity before they had the basics right. Chasing agentic workflows and automated pipelines before they'd established what problem they were solving.

For smaller marketing teams, this is actually good news. You don't have the budget to make the expensive mistakes. And constraint, used well, is a superpower.

A Simple Five-step Approach to Getting AI Right

Here's a five-step approach that works — manageable, low risk, and built to last.

Step 1: Start with a pain point, not a technology

Before touching a single tool, write down the three most repetitive, time-consuming tasks your marketing team does every week. First drafts of social copy. 

Summarising competitor content. Pulling together campaign reports. These are your entry points.

AI delivers fastest and most measurably when it's solving a specific, recurring problem. Pick one task, run a two-week trial with a simple tool like Claude or ChatGPT, and measure the time saved. That number becomes your business case for everything that follows.

Step 2: Build prompting skills before you build pipelines

The single biggest reason AI tools underdeliver is that people treat them like a search engine. You get real value when you learn to write clear, detailed prompts.

Run internal prompt challenges where team members share what worked and what didn't. This costs nothing and builds the collective knowledge that makes every subsequent AI investment more effective.

Step 3: Adopt one platform and go deep

Resist the temptation to trial five tools at once. Pick one AI platform and use it daily across your team for 60 days.

AI rewards familiarity. Teams using the same tool consistently discover use cases they never anticipated at the start. A shared workspace where prompts and outputs are saved means you start building reusable templates — and that's the foundation of everything that comes next.

Step 4: Introduce automation in one contained workflow

Once your team is comfortable with AI-assisted writing and research, identify one workflow where a simple automation would save meaningful time.

A good starting point: a new blog post is drafted, AI summarises it into three social posts and a newsletter intro, and these are saved to a shared folder for review. Tools like Zapier or Make connect your AI platform to the apps you already use, no code required. Keep the human in the loop at every output stage. The goal is to remove friction, not oversight.

Step 5: Evaluate, document, then expand

After 90 days, hold a retrospective. What saved time? What produced poor output? What needs a human correction every time?

Document this honestly. It becomes the foundation of a responsible agentic strategy — because agentic tools, where AI manages sequences of actions autonomously, multiply both your gains and your mistakes. Teams that expand into agentic workflows with clear quality standards and human review triggers are the ones that scale successfully. Those that rush in without this groundwork are the ones that become a statistic.

The Bigger Picture

The bigger picture is simple. The teams that will look back on this period with satisfaction aren't the ones who moved fastest. They're the ones who built carefully — who started small, learned quickly, and expanded only when they'd earned the confidence to do so.

If that's where you are — ready to stop experimenting and start building something that lasts — we can help.