We are increasingly asked how we use AI. That makes sense, because almost every organisation is working with it in some way. Some are experimenting with prompts. Others are building agents, workflows or full automations. But in the end, the question is not what AI can do. The real question is what it adds.

Because AI can speed up a lot. But what we see is this: AI accelerates many parts of the process, while taste, experience and judgement remain human.

AI only becomes valuable when you know what you are working towards. If you have a clear goal, AI can help you get there. But when you do not know what you want to achieve, you get output that may sound convincing but does not make sense. Whether a text is sharp enough, whether a structure feels logical, or whether a visual truly explains what it needs to explain, that always requires our judgement.

We mainly use AI to understand faster

Many of our clients work with complex subject matter. Years of research, specialist knowledge and technology that feels obvious to them, but not to an investor, customer or partner. That is where we use AI.

Not to automatically write a story for us, but to get to the core faster. To ask better questions, see connections and make complex information easier to understand.

Good output needs good input and a lot of context

What we see again and again is that the more specific you are about what you need, the better the output becomes. It also becomes easier to judge whether the output is right.

AI only works well when you feed it properly. Who are you writing for? What is the goal? Which platform is it for? What should the reader think, feel or do afterwards?

One thing we often do is ask AI to ask us questions first, instead of giving an answer straight away. “Ask me five questions that will help you understand this better.” That approach almost always leads to better output.

And once the output is there, keep asking. “How did you arrive at this answer?” helps you understand what sits behind it. It also makes you a little better at working with AI every time.

A small step every day, not one big workflow

Many organisations start with enthusiasm. An agent is built, a workflow is set up, a pilot is launched. But a few months later, little is still being used.

Not because AI has no value, but because adoption is difficult. Starting too big makes it overwhelming.

For us, it works because we start small and keep improving. A little better every day. A text made sharper. A structure tested more quickly. A question formulated more clearly. Step by step, AI stops being a separate experiment and becomes a natural part of the way you work.

Human judgement remains essential

The risk starts when you lean back. When you give one simple prompt, stop checking the output and assume it will be correct.

That is when quality drops quickly.

We have been doing this work for twenty years. We know where small nuances can make a big difference. That experience cannot be replaced by a prompt. AI can help us think faster, but only if we continue to look critically ourselves.

What you can take from this

Do not start with the question: which AI tool should we use?

Start with the question: where would the quality of our work improve if we had more time to spend on it?

Or: what part of our work is so repetitive and time-consuming that it costs energy without adding real value?

Choose one concrete area. Use AI there every day. Give context, ask the right questions and keep judging the output yourself.

Because AI speeds up a lot. But taste, experience and judgement remain human.

And those three ultimately determine whether the work gets better.