Insights

Your agency is already using AI on your brand. Time to do it properly in-house.

Someone is already using AI to make your content. The question isn’t whether AI touches your brand, it’s whether you’re the one setting the rules for how it sounds.
26 Sep 2026
AI, Marketing
A speech bubble built from dark toy bricks between two small stacks of bricks, with a blue quotation mark being added on top.

Bring the high-volume, routine content in-house and pay properly for the original thinking. Write the brand voice down in a form a model can follow, feed it your own material, and keep a named person accountable for what ships.

Here’s a safe assumption. The agency producing your content is already using AI to help make it. Not instead of their people, but alongside them, to draft, to generate variants, to speed up the parts that used to take days. Most won’t lead with it, and that’s fair enough. You’re paying for the result, not the method.

But it changes the question in front of you. AI is touching your brand whether you planned for it or not. So the real decision isn’t whether to allow AI into your content. It’s whether you want to keep paying agency rates for work you could increasingly do in-house, and whether you’d rather be the one deciding how your brand sounds when a model is involved.

For a lot of routine content, the answer is now clearly yes to bringing it in-house. The interesting part is how you do that without your brand voice dissolving into the same flavourless output everyone else is generating.

This will ruffle some feathers, so let me be straight

I know how this sounds, and I know some of my creative friends won’t love it. So let me be clear about what I’m actually saying, because it isn’t “AI can replace good creative people”.

It can’t, and the opposite is closer to the truth. As routine production gets cheap, original thinking gets more valuable, not less. The strategist who cracks the real problem, the writer with a genuine voice, the art director with taste, the agency that brings a perspective you can’t get from inside your own building: all of that is worth more in a world drowning in generated content, not less. That work I’d pay proper money for and protect.

What’s changing is narrower, and it’s two things. First, the high-volume, repeatable production, the fortieth variant and the ninth localisation, is work a model now does in minutes, and paying craft rates for it is getting harder to justify. Second, the pretence that AI isn’t already in the process helps nobody. The honest agencies are already having this conversation with their clients. The ones who should be nervous are the ones billing by the day for work the day no longer requires.

So this isn’t a case against agencies or against creative people. It’s a case for being honest about which work is which, doing the routine part in-house where it now belongs, and paying properly for the real creative partnership. If that annoys anyone, it’s probably worth a conversation rather than a quiet shelving.

Should you bring AI content in-house or leave it with the agency?

Bring the high-volume, repeatable work in-house and keep the agency for the work that genuinely needs their craft and outside perspective. AI has made first-draft copy, social variants, localisation and image generation fast and cheap enough that a lean in-house team can now do work it used to outsource. That frees the agency relationship for strategy, big creative ideas and the things AI is bad at. The dividing line isn’t in-house versus agency any more. It’s routine production versus original thinking.

The teams getting real value from this aren’t the ones who bought the newest model. They’re the ones who built a proper process around it. The pattern across the winners is the same everywhere: AI works best inside a governed workflow, not as a bolt-on to whatever tool is having its moment.

The brand voice problem, and why it's solvable

The fear people have is reasonable. Bring content generation in-house, and everything starts to sound like AI. Bland, over-eager, weirdly keen on the word “elevate”. Your brand voice, built over years, quietly replaced by the house style of a language model.

It happens. But it happens to teams who type a request into a generic tool and paste the result. It doesn’t have to happen to you, because a brand voice can be written down in a form a model can actually follow.

That’s the move that changes everything. Not a vague brand book that says you’re “bold and human”, which means nothing to a machine or, frankly, to a new copywriter. A specific, worked set of rules: the words you use and the ones you never use, how long your sentences run, whether you use contractions, what you sound like when you’re explaining something hard, with real before-and-after examples of your voice done right and done wrong. Give a model that, and its output starts sounding like you rather than like everything else. Withhold it, and you get the average of the internet.

I know this works because I run my own writing that way. The rules are written down, the banned phrases are listed, the examples are there, and anything drafted with AI gets checked against them before it goes anywhere. The voice survives the tool. That’s the standard to aim for.

Learn from the ones who did it at scale

Two examples worth knowing, because they show the range.

WPP built an AI operating system with a content engine, and for one client trained custom models on that brand’s characters, so brand-character content that used to take months could be produced in minutes, on-model and reactive to what was happening that week. That’s the industrial end: the agency itself reinventing how creative gets made. You don’t need their scale to take the lesson, which is that the value came from training the AI on the brand’s own material and wrapping a real process around it, not from the raw model.

Coca-Cola went the other way and opened it up. Their “Create Real Magic” work let fans co-create branded visuals using generative AI trained on the company’s creative archive. It turned the audience into participants and built engagement without inflating the budget, precisely because the AI was grounded in decades of genuine brand assets rather than generating from nothing.

Different strategies, same foundation. Both trained the AI on their own material, and both kept a firm hand on what shipped. Neither just switched on a generic tool and hoped.

How to set it up without losing what makes you you

If you’re bringing content and creative in-house, a few things need to be true from the start.

Write the brand voice down properly, in the specific form described above, and treat it as a living document that gets sharper every time the AI gets something wrong. Feed the AI your own material, your best past work, your real customer language, your actual products, so it’s generating from your world and not the generic one. Keep a person accountable for what ships, with the time and the standing to send work back. Decide what you’ll never let AI do on its own, even if it could, because some things carry too much of the brand to hand over. And be transparent where it matters, so you stay the right side of the trust line with your audience.

Get that right and the economics are hard to argue with. You produce more, faster, at lower cost, and it still sounds like you. Get it wrong and you’ve simply automated the production of content nobody can tell apart from anyone else’s.

What to do this week

Find one type of content you currently outsource that’s high-volume and fairly routine. Social variants, product descriptions, standard email copy. Then ask what it would take to bring it in-house well: the voice rules written down, the right material to train on, and a named person to own the output. Run it as a small, honest test against what the agency currently gives you, on both quality and cost.

You’ll learn quickly whether it’s worth expanding, and you’ll have kept your brand voice in your own hands while doing it.

Already making content with AI?

If your content is already being made with AI and you’d rather set it up properly in-house, with the brand and tone of voice protected, I can show you how. Ping me a message or let’s grab a coffee.

Share this insight

Related insights
A faint grid of binary zeros and ones on cream, with a single blue magnifying glass to the right.
AI, Digital

Scotland’s biggest law firms have cracked SEO. Almost none are ready for AI search.

Scotland’s biggest law firms have cracked SEO, but almost none are ready for AI search. What we found across the ten largest firms, and the five fixes that move a firm from invisible to quotable.
2 Oct 2026
Three nested rounded frames on a cream ground, scattered with dark dots and rings, a few of them picked out in blue.
Marketing, Risk and governance

The riskiest AI in your marketing team is the tool nobody approved

How marketing teams can use AI without breaking data-protection rules or brand trust. Right-size the oversight, name an owner, and start with the tools already in use.
29 Sep 2026
Many tangled dark lines, two of them blue, drawn together from the left into one straight line that ends in a single dot.
AI, Marketing

Before AI optimises your campaigns, fix what you’re measuring

AI can optimise campaigns and share credit across channels, but only if the data underneath is joined up and honest. How to get attribution AI can actually use.
28 Sep 2026

Let's grab a coffee.

If something's stuck, or you'd like a second opinion on a plan, tell me about it. No hard sell.