AI is the icing on the cake – but first comes the homework

Michael Schmidtke, Vice President Content Flow and Digital Channels at Bosch on content flow, virtual influencers and data-oriented communication at Bosch

AG CommTech: Michael, Bosch is known in the communications industry for creative cases and innovation awards. Please give us an overview: How are you positioned in communications?

Michael Schmidtke: We are basically organized according to target groups: There are teams for internal communication, media relations, digital communication, customer and brand communication and political communication. Some of our colleagues work in Germany and others in the regions and countries. The basis for data-oriented communication is the so-called content flow process – a cross-divisional process in which content creation, distribution, analysis and optimization are seamlessly interlinked.

AG CommTech: You deliberately decided against a classic newsroom model. Why?

Michael Schmidtke: Because we see communication as a networked process. We don’t make a strict distinction between content and channel. For us, three components count: Technology, content and analytics. We bring them together in terms of processes – across departments, efficiently and practically.

AG CommTech: What role does artificial intelligence play in your communications department?

Michael Schmidtke: We have been working with AI since 2017 – long before ChatGPT and co. We have now launched a cross-divisional project to systematically check where AI can support us. This is particularly visible in the visual area: for example, we are testing new formats with chatbots or “Sparky”, a virtual influencer on Instagram. AI also helps us with content creation, translations and localization.

AG CommTech: What can we imagine by “Sparky”?

Michael Schmidtke: “Sparky” is an animated historical spark plug that appears as a virtual character. “Sparky interviews Bosch experts or reports from events such as CES – from inside an AI refrigerator, for example. Such projects would not be possible in this form without AI. Especially in the moving image sector, where channels need constant output to remain relevant, AI gives us new scope – also with a view to internationalization.

AG CommTech: Do you also use a chatbot for internal communication?

Michael Schmidtke: Exactly. “BeeGee” – named after our Bosch Global Net (BGN) – is a chatbot that supports over 10,000 intranet editors. In the past, hotline inquiries were made or manuals leafed through. Today, “BeeGee” provides precise answers in a matter of seconds. This has drastically reduced the number of support tickets – a real efficiency gain.

AG CommTech: And in media analysis – how do you use AI there?

Michael Schmidtke: Our service providers have been using classic AI methods for years, for example for semi-automated tonality assessment. What’s new is that we are now using generative AI ourselves, for example to support situation reports. Individual simple articles, for example, can be summarized quite well with the help of AI. However, human review remains important. When it comes to more complex topics, the limits of today’s AI quickly become apparent, for example when several articles on a topic need to be condensed into a meaningful summary.

AG CommTech: Is media analytics still up to date – or is an AI-supported management briefing enough?

Michael Schmidtke: We already had situation reports before AI. The difference is that we now have an additional tool to make the existing content even more efficient. But without a clean content database, clearly defined processes and a critical team, none of this will help. Just because generative AI delivers initial text drafts does not replace human judgment in complex contexts. Anyone who believes that AI can simply be given free rein will soon be disappointed.

AG CommTech: What exactly is your content flow?

Michael Schmidtke: This is our cross-divisional process: defining topics, developing content, playing it out, measuring its impact, optimizing it. And we do this in close cooperation with analytics and creation. We evaluate at output and outcome level – including target group and impact analyses. The goal: data-based decision templates. AI helps us to become more efficient – but the process comes first and forms the decisive basis.

AG CommTech: How do you process real-time information with the help of AI?

Michael Schmidtke: One keyword (buzzword!) that we are working on is “agentic AI”. We are experimenting with corresponding models in which specialized AI agents – for example, as described above for article synopsis – work together in workflows. It works like a one-two in soccer: one agent summarizes individual articles, the next summarizes the results. Transparency is important to us – every AI action must be traceable.

AG CommTech: That sounds technically demanding. How do you enable your team to do this?

Michael Schmidtke: We are focusing on hybrid roles – in other words, specialists at the interface between communication and technology. I don’t believe that all communicators will need a “black belt in prompting” in the future. We will remain creative content people. But we need colleagues who can align AI tools in such a way that they offer maximum added value specifically for our communication applications.

AG CommTech: And what advice would you give to those 46 percent of communications departments that don’t even measure tonality?

Michael Schmidtke:
Start with the basics. Understand your target groups, define your processes, structure your data. Only when this is in place does the use of AI make sense. It’s the icing on the cake – but without a solid foundation, it has no effect. If you proceed in this way, you will notice where AI can provide support along the way: more efficient, cheaper, more scalable. But not technology first – target group first.