We recently finished teaching our first full cohort of the Applied AI Certification course for the In-House Agency Forum (IHAF), the only professional association dedicated exclusively to in-house advertising and creative teams. Let me set the stage: four sessions and sixteen hours with 26 creative and marketing leaders from 16 in-house agencies across healthcare, finance, insurance, consumer goods, and more.
We came in to teach and share what we’ve learned over the past three years on our AI journey. But what stayed with us most wasn’t what we taught. It was what we heard.
The conversations happening inside creative teams right now are far more honest, nuanced, and human than most of the AI discourse happening online.
So consider this a field report from the front lines of AI adoption.
1. Everyone is in it.
We came in expecting a wide range. Some teams are far along, some at zero, most somewhere in the middle.
That’s not what we found.
The work was already broad and nobody was starting from zero: summarizing transcripts, refining copy, generating and extending imagery, adapting content across channels, synthesizing research into strategy, even building custom GPTs to pressure-test audience thinking.
Across the cohort, self-rated AI value averaged 3.65 out of 5.
The dominant industry narrative right now is “We need to start using AI.” But that’s not what we found inside these teams. They’ve already started. In many cases, they started a while ago.
The question is no longer whether AI is being used. The question is what to do about the fact that it’s already embedded inside the work.
That’s a very different starting point. And it changes what useful leadership actually looks like.
2. The problem isn’t adoption. It’s visibility.
If everyone is already using AI, you’d expect that to feel obvious inside an organization.
It doesn’t. And that gap may be the single biggest source of stalled momentum we observed across the cohort.
Again and again, we saw real AI value being generated quietly by individuals inside workflows nobody else could see. One person was getting tremendous results using Claude for transcript review while the colleague sitting next to them had no idea it was even possible. One designer had developed a generative-fill workflow that saved hours per asset while teammates were still doing it manually.
Participants used remarkably similar language to describe the state of AI inside their organizations: “ad hoc,” “individual,” “inconsistent.”
The wins are already there. They’re just trapped inside private practice.
The teams making the most progress aren’t necessarily the ones chasing the newest tool. They’re the ones surfacing the work their people are already doing, comparing notes openly, and turning isolated experiments into shared capability.
The opportunity isn’t future AI. It’s the AI already happening inside your building.
3. The brief is broken.
When we asked participants to map the biggest sources of friction in their creative production process, one answer dwarfed the rest.
Sixty-one percent named poor or incomplete briefs as the number one problem. Higher than capacity. Higher than scope creep. Higher than review bottlenecks. Higher than tooling.
What struck us wasn’t just the percentage. It was the universality.
The same frustrations surfaced across financial services, healthcare, insurance, food, and consumer goods — organizations with almost nothing else in common. Briefs arrived missing audience, goals, or context. Stakeholders struggled to articulate what they actually wanted. Creative teams spent enormous amounts of time chasing information that should have existed from the start.
One participant described the breakdown so completely that briefs had almost disappeared entirely:
“We use briefs in maybe 5% of work. The rest is conversations with PMs trying to figure out what’s actually being asked.”
Most AI conversations focus downstream: faster output, more variations, accelerated production. The cohort kept pointing upstream.
The highest-leverage place to apply AI inside an in-house agency may not be the asset. It may be the intake. Because if the brief is broken, AI just helps you produce sharper-looking versions of the wrong thing, faster.
4. Look for the helpers.
One of the most useful patterns we observed across the sessions was this: the help that actually moves teams forward comes from two places, and most organizations are underusing both.
The first is AI itself — not as a production engine, but as a thinking partner.
The participants getting the most value from AI weren’t necessarily generating the most output. They were using it to challenge assumptions, pressure-test thinking, reorganize ideas, and uncover angles they hadn’t considered.
One participant described it this way:
“Sometimes it challenges me to organize my communication differently. Occasionally it suggests something better than what I originally proposed. Humbling.”
That posture produced consistently stronger work than treating AI like a junior employee waiting for instructions.
The second helper was each other.
Some of the most valuable moments in the cohort weren’t anything we taught. They were moments when leaders from completely different organizations realized they were facing the exact same wall — and one of them had already found a way through it.
A compliance question that someone in healthcare had been stuck on for months had quietly been solved by someone in insurance. A custom-GPT workflow that a financial services team was hesitant to try was already running successfully inside a CPG organization.
Most in-house leaders are trying to figure this out alone inside organizations where peers are too busy, too cautious, or too far away.
They shouldn’t be. The help they need is mostly already out there. They just haven’t been in the same room with it yet.
5. Humans matter more than ever.
The quietest pattern in the cohort was also the deepest.
It rarely surfaced first in group discussion. It almost always appeared privately in intake forms or side conversations near the end of a session.
And it wasn’t really “Will AI replace me?”
It was something more nuanced than that.
How does the non-humanness of AI align with a brand built on trust? What happens to craft when AI becomes part of the workflow? What skills begin to atrophy? What happens when clients or colleagues start expecting “AI speed” without understanding the strategic and creative judgment still required underneath the surface?
One participant described the tension painfully well:
“A colleague submits an AI-generated output and asks me to make ‘that,’ and I have to explain why their want won’t work for the actual design need.”
That’s the real question underneath all the others. In a world where almost anyone can generate a confident-looking first draft of almost anything, what makes the work coming out of your team your work?
The answer isn’t less AI. It’s more human.
More taste. More point of view. More judgment about when to use AI and when to put it down. More ability to recognize confident-but-generic output and push past it toward something original, strategic, and emotionally resonant.
AI raises the floor on output quality everywhere. That means that the ceiling — the part requiring craft, conviction, and distinctly human perspective — becomes the only place where real differentiation still lives.
The leaders doing the strongest work right now aren’t the ones using AI the most.
They’re the ones getting clearer about what only humans should do — and protecting that ground fiercely while letting AI carry the rest.
That’s not a question with a clean answer.
But it’s the right question.
And in some ways, it may be the only one that matters.
So where does that leave you?
The organizations winning with AI right now aren’t automating creativity away.
They’re getting brutally clear about where human judgment matters most — and redesigning their teams around protecting it.
The teams adapting fastest aren’t the ones chasing every new tool release. They’re the ones creating cultures where people can learn together, experiment openly, share what’s working, and push each other toward better thinking and better work.
That’s exactly what our AI training programs are designed to help organizations do.
If your team is ready to move beyond surface-level experimentation and build an AI approach grounded in real creative workflows, we’d love to build a bespoke program with you.
