Building AI Competence Within a Team: What Really Works
In brief: Building AI competence in a team is not a matter of finding the right tool — it is a matter of putting the right structure in place. Teams that want to accelerate AI adoption need clear guidelines, differentiated learning paths by level, and above all, concrete use cases that show what is genuinely possible. At our third Raw & Roasted pre-work event on June 25, 2026, our colleague Dr. Özgün Köksal (Senior Data Analyst & PhD in Learning Sciences) discussed exactly these questions with our audience.
From corporate employees to freelancers, from marketing professionals and BI specialists to corporate compliance: the audience at the third Raw & Roasted in our office at Landshuter Allee 49 was more diverse than ever. What became visible there also shows up in our consulting work: AI is not an IT topic and not a marketing topic. It is a cross-functional topic that affects everyone, and one in which everyone still has a lot to learn, across industries and hierarchies. We may even be living through one of the steepest learning curves of our working lives.
Özgün put one question at the centre of her talk: How can a team use AI in ways that genuinely help it achieve its shared goals?
Same Team. Very Different Starting Points.
Almost everyone in the room described the same thing: in their own team, some people experiment with AI every day, while others avoid the topic. Sometimes openly, sometimes quietly. And it has little to do with age or experience. We have seen senior leaders and recent graduates on both sides.
What lies behind the reluctance is usually not a lack of motivation, but a lack of orientation. Those who do not know where to start, do not start. And those who quietly step back lose ground over months without anyone noticing. That is the real risk: AI competence that is not consciously developed spreads unevenly across the team and generates no shared momentum.
Özgün showed in her talk the four typical AI starting points that appear in almost every team:
An important point from the talk that landed clearly in the room: these are not fixed types. The same person can be in a completely different place depending on the task. Someone who writes texts with AI every day may be just starting out when it comes to data analysis with AI.
The Guidelines Paradox
One topic that really got the discussion going: AI guidelines. Many organisations in the room already have them. But where are they? Usually somewhere in the intranet, hard to find, sometimes so extensive that nobody reads or fully understands them.
Many employees had already integrated AI into their daily work long before the first rules arrived. When guidelines then retrospectively prohibit certain things, frustration follows. And in the worst case, people shift to personal devices and do privately what is not permitted on company hardware. That serves neither data security nor company culture.
The conclusion from our discussion was clear: guidelines need to be as short as possible. They should create orientation, not slow down creativity. What a sensible AI guideline needs to achieve can be reduced to three categories: what is clearly permitted, what needs alignment, and what is off-limits. More often than not, that is enough to start with.
Too Many Scouts, Too Little Direction
Özgün also introduced the concept of the AI Scout: one or two people on the team who test new AI tools, filter them, and translate what they find for the rest. That prompted direct responses from the audience, because in some teams there are simply too many Scouts. The result is not more orientation, but more noise. Too much input, too many parallel experiments, no shared thread. Many organisations are still finding the right balance here.
The real strength of a Scout lies in the translation work: is this tool or AI process actually relevant for us? Does it solve a real problem? Is it ready for everyday use? This filtering effort is underestimated. And it only works when the Scout stays close to the team’s actual workflows.
The Super-Power for AI Motivation: The “I Want That Too” Moment
What genuinely moves AI adoption in practice is not another workshop. It is the moment when someone sees what is possible and thinks: I want that too.
Concrete examples came from the room. Someone shows how a formatting task that used to take hours now takes a few minutes. Another person describes how weekly campaign reports that were previously assembled manually are now automated. These moments often have more impact than a big, well-announced pilot project, because they are immediately relatable for everyone.
Knowledge sharing sessions that make these everyday wins visible emerged as one of the most effective levers in the discussion. Particularly when they happen among people with similar tasks. Those who do the same kind of work learn from each other faster and more concretely than in a large mixed group.
The Real Challenge: Not the Tool Question, but the Right Starting Question
Something we recognise from consulting work, and that the discussion confirmed again: the most common reason AI initiatives do not achieve the hoped-for impact is not a bad tool. It is the missing starting question.
Many teams start with the tool selection before it is clear what problem needs to be solved. The more useful order is the reverse: where are we losing the most time today? Which tasks are recurring and time-consuming? Only once that is clear does the tool choice make sense. And only then do use cases emerge that actually change something.
Four Practices for Building AI Competence in Teams
Özgün presented four concrete practices that help teams build AI competence systematically:
1. Plan AI learning by level, not by team. Not everyone needs the same thing. A baseline level for everyone, applied knowledge for most, deep methodological expertise for one or two people. This differentiation matters because it acknowledges different starting points rather than pretending everyone is at the same stage.
2. Establish the AI Scout, but keep it focused. Who explores what is new? Who filters what is actually relevant? Who translates new possibilities into concrete use cases for the team? A Scout who fulfils all three functions moves the team forward. Multiple Scouts without coordination bring more input, but no direction.
3. Create usable guidelines. Short, clear, easy to find. Three categories are enough: yes, maybe, no. What teams do not need are multi-page documents nobody reads. What they need is orientation that enables fast decisions.
4. Start with a real piece of daily work. Not a greenfield pilot project, but a concrete task that costs time today. That produces visible results quickly and makes AI tangible for everyone.
Dr. Özgün Köksal
Senior Data Analytics Specialist, Hopmann Marketing Analytics
Özgün advises organisations on the adoption of AI in marketing, building analytical capabilities, and developing data-driven decision structures. Her work combines methodological depth with an understanding of how organisations actually function.
What’s Coming Next
This morning showed again what the Raw & Roasted format is about: not a frontal presentation, but genuine exchange between people working through the same questions. Anyone who wants to join the next session: on 23 July 2026, our Marketing Science Manager Dr. Simon Hannemann will give a practical introduction to Marketing Mix Modeling (MMM) — what it delivers, where its limits lie, and when it is worth it.
For those who would rather explore the topic of AI in a larger setting: on 28 July 2026, our event AI in Practice takes place, with a keynote and interactive sessions on AI Search Visibility (GEO), Agentic AI, Marketing Mix Modeling, and AI competence. For marketing, sales, and IT decision-makers from mid-sized businesses. This is a partner event with the German Association for Medium-Sized Businesses (BVMW).
Raw & Roasted | 23 July 2026, 8:30 AM
23 July 2026 · Munich · Hopmann Office, Landshuter Allee 49
Marketing Mix Modeling explained in practical terms. Dr. Simon Hannemann gives a hands-on overview: what MMM delivers, where its limits lie, and when it is worth it. Inspiring talk, open discussion, and barista coffee.
Details & Registration
AI in Practice | 28 July 2026, 5:00 PM
28 July 2026 · Munich
Keynote and interactive sessions on AI Search Visibility (GEO), Agentic AI, Marketing Mix Modeling, and AI competence. For marketing, sales, and IT decision-makers from mid-sized businesses.
Details & RegistrationAll upcoming dates are available in our events overview. Anyone who wants to go deeper on the topic: the blog post AI Evening: What Marketing Managers Need to Know covers further perspectives from our BVMW evening in May.
Get in touch with HopmannWhat our clients often want to know.
FAQ: Building AI Competence in Teams
How do I build AI competence in a marketing team?
The most effective starting point is not a tool training, but the question: where are we losing the most time today? From there, concrete use cases emerge. Combined with differentiated learning paths by competence level and clearly communicated guidelines, this creates a structure that drives sustainable AI adoption.
What is AI literacy and why does it matter?
AI literacy is the ability to use AI tools effectively, evaluate their outputs critically, and understand where AI creates genuine value in a specific work context. It is not an individual skill — it is a team capability. A team with high AI literacy collectively makes better decisions about where and how to apply AI.
How should AI guidelines in a company be structured?
As short as possible, as clear as necessary. Three categories are enough to start: what is clearly permitted, what requires alignment, and what is off-limits. Guidelines that are too extensive are not read and can push employees to use AI privately outside company systems. Orientation beats regulation.
What makes a good AI Scout in a team?
An AI Scout explores new tools, filters what is worth the team’s attention, and translates new capabilities into concrete use cases. The translation work is what matters most: not every new tool is useful, and not every capability matches the team’s current level of readiness. More than two Scouts per team tends to create noise rather than progress.
Why do many AI initiatives in companies fail?
The most common reason is not the wrong tool, but the missing starting question. Many initiatives begin with the tool selection before it is clear what problem needs to be solved. AI delivers value when it connects to specific, time-consuming tasks, not to abstract transformation goals.
How can you speed up AI adoption in a team?
Concrete use cases that are made visible work better than any training programme. Knowledge sharing sessions — where colleagues show how they have simplified a specific task with AI — create the decisive moment: “I want that too.” Particularly effective among people with similar roles and responsibilities.