The Small In-House Advantages That Turn Out to Be Huge
People talk about AI as if unlocking its value requires a research lab. It doesn't. The real insight is actually pretty mundane: most people using AI are lazy about it. And yes — that includes me. If I'm honest about my own prompts, they're often two or three lines long. That's a poor brief. It would be a poor brief if I handed it to a person, too. "Can you just write something for this campaign? Make it sharp." That's not a brief. That's a wish list. So when someone says AI isn't impressive, it's often because they fed it the equivalent of telling a chef: "Make something delicious."
It's not rocket science. It's structure.
We have three copywriters. We have a lot of car brands. Every brand has its own tone. Its own rules. Words we're allowed to use. Words we're not. Global guidelines. Local quirks. Legal constraints. You could write a long, detailed prompt every single time:
Which brand?
Which universe?
Top-of-funnel or tactical?
Which media?
Which phrasings should be avoided?
What do we need to pay special attention to?
You could.
But you don't.
Because you're busy.
So we did something about it. And this is just one example of how I think AI will change day-to-day life for agencies, marketing departments, in-house teams. And, really, everyone else.
Wordsmith: a proper brief, built into the system
Wordsmith didn't come about because we invented new technology. It came about because we admitted we couldn't be bothered writing long, carefully worked-out briefs every time. So we standardised it. In practice, our ZITE Wordsmith is just a simplified agentic workflow.
First, it identifies the context:
What are we writing?
Which brand?
Which funnel step?
Which universe?
Then it pulls in:
Brand tone of voice
Formal requirements
Funnel logic
Media-specific considerations
Then one AI writes the copy.
But we don't stop there. A second AI reads the text afterwards and assesses:
Has the task actually been solved?
Does it hit the brand?
Are there phrasings that should be adjusted?
And, not least: why?
It makes the case for its own suggestions.
The copywriter takes it from there. It's human in the loop. It's still the human eye, relevance and judgement the client is paying for.
But we've removed the part where you sit staring at the screen thinking, "Did I remember everything?" It's not magic. It's a proper brief, built into the system.
The most boring task in the world
At ZITE, Google Ads copy still gets delivered in Excel. It might be the single most uninspiring place in the world to write copy. So mistakes creep in. Double spaces. Odd line breaks. Small typos. Columns that don't quite line up.
And the truth is: it's the small mistakes that irritate people most. Not the big blunders. The small, mechanical ones. The kind that make you think, "Could they not just have read it through before sending it?"
We could see it was one of the places where we made the most mistakes. Not because we didn't know better. But because the task is mechanical and dull. So we built a checker. You upload the document. It reviews it. Finds the errors. Hands it back. It took a couple of hours to build. Genuinely not rocket science.
But we haven't had a single error in a Google Ads document since November. And that's actually a pretty good feeling.
The real advantage of sitting in-house
We can do this because we sit close to the business. The data stays with us. The brand knowledge stays with us. The context is built in. We don't have to explain our own tone of voice every time. We can build it into the system. We can adjust the same day. Iterate without sending briefs back and forth. Fine-tune, because we feel it in our day-to-day work. This isn't a transformation project. And it isn't a group strategy.
It's looking at your biggest bottlenecks, sources of error or friction, and asking: can we build something small that solves exactly this?
AI doesn't replace people. It replaces friction.
The problem isn't the technology. The problem is that we use it half-heartedly. A two-line prompt. Half a brief. Full expectations. When you build the structure into the system instead, you get quality. Not because AI is brilliant. But because the brief becomes brilliant — or at least a lot better.
And that's where the small in-house advantage becomes huge. Not because we invented something revolutionary. But because we got our act together and built something small.
