We know the industry.
And the access to its decision-makers.

The task

A business unit of a chemical group needed a white paper for several target regions. Not a classic technical white paper with problem, data and process, but a text on the state of an entire segment: geopolitical shifts, flexibility under changing conditions, the question of how to keep supply and service stable for customers. That is really e-book material. Here it had to fit on about ten pages without turning into a brochure, and deliver leads that reflect the quality of the topic.

Our approach

The client provided the brief, the material and the tone of voice. We process such input with AI, primarily with one model trained on our methodology over months, complemented by other models for research and cross-checking. The AI only accesses public sources; everything else comes from the client. And it is a one-way street: client data never leaves the process and is used for nothing else.

The first draft followed the brief exactly. It was clean, and it was too smooth. The difficulty the topic had from the start was back in the text: general, descriptive, without an edge. We told the client openly and asked for a second attempt, this time with our own method: straight, fact-based, no embellishment, with far more sources. Several iterations before the draft even reached the client, all within two days.

The second draft hit the mark. The third was only the client’s feedback worked in. No draft goes out without these review loops with the experts on the client side; we read every document in full before anyone else sees it.

The result

The paper went through the specialist departments and unit management and was approved without objections. The feedback: on the level the team would have aimed for internally, except that the time would not have been there. Three years ago a paper like this cost 30 to 40 working hours with a ghostwriter and expert alignment. Today net working time is about one day; calendar time depends on the reviewers’ availability. The AI now also handles the corporate design, so only individual wishes remain to be adjusted at the end.

What we take from it: quality does not come from “AI, write me a white paper”. It comes from a prompt and a process built up over a long time, from research you check yourself, and from experts who take responsibility for what goes out.

The task

IMP was founded in 1976 as a pure publisher representation for the German market, extended to international titles in the 1980s and consolidated in the 1990s when steel and mechanical engineering fell away as revenue carriers. Between 1997 and 2003 the first service of our own arrived: competitive overviews of trade media, still copied by hand from printed media guides into Excel. In 2003 came the step to an agency with media planning, negotiation and handling, with the first budget from Degussa and the deliberate decision for the chemical industry as the basis of all other industries.

Until early 2023 that ran stably. Then large language models became broadly available while advertising budgets fell under the pressure of the European economy. Together that meant: the offer had to change again, and faster than in 2003.

Our approach

The impulse came from inside: Kyra Lutterman pushed us not to watch AI but to use it. The first attempts with modular tools were not enough. So we worked our way into the models themselves and tested them on real client projects: e-books, white papers, LinkedIn campaigns. 2024 and 2025 brought the plateau. Internal bans applied in many companies; AI was used for emails and summaries, rarely for more. As an owner-managed company we could judge the risk ourselves and invest time and money, including in private projects whose lessons flowed back into IMP. Mistakes were part of the plan, within a range that stays controllable.

Since late 2025 we work with several models in parallel, because each has its strengths, and with prompts and processes that have grown over years. What we learned for ourselves we apply for clients: we monitor the media landscape and feed changes into planning. For a client with a rigorous KPI logic we built the evaluation so that it drops straight into his internal report. For key accounts we produce a communication analysis from the outside: what is publicly visible, where the strengths are, where the gaps, as the basis for the next campaign.

The result

IMP today is more than media planning. We look into markets before we propose platforms and bring market picture, client goals and budget into line. Out of that came the Intelligence Series, the Media Intelligence Reports and the communication analyses. AI has taken business from us and brought business to us; the balance is positive because we did not wait out the transition. What remains: nothing works without the human component. AI accelerates, extends and asks questions. Responsibility for what goes out stays with us.