My Key Insights from the Conversation on the CIO Podcast “Inside IT” with Florian Mertl
Recently I had the honor of being a guest on Florian Mertl’s renowned CIO podcast “Inside IT.” It was a refreshingly honest exchange. Far from the usual marketing hype that often dominates the AI discussion. We looked deep into practice: Where does the German SME sector really stand? What are the real hurdles in AI adoption? And most importantly: How do you overcome them?
The conversation confirmed many of our daily experiences at innFactory. The biggest challenge in AI projects is rarely the technology itself. The real hurdle is the last mile to the employee and the often neglected technical homework.
For those who want to listen in directly, here’s the full episode:
For everyone else, I’ve summarized the four central theses from our conversation that are crucial for every company on the path to AI integration.
1. The Adoption Dilemma: A Tool Is Not Yet a Strategy
“Just because I now have a Company GPT doesn’t mean it will be used. Probably 90% of the workforce won’t use the tool in the end.”
This is not pessimism, but a realistic observation. Many companies believe that buying an expensive license (e.g., Copilot for everyone) is the starting shot for AI transformation. In reality, it’s often just the beginning of an expensive misunderstanding.
Successful AI adoption is not an IT project that you dump in a department. It’s a holistic change process that must be anchored at the executive level. At innFactory, we use our “Adoption Ladder” for this:
- Create awareness: Train management and employees. What is AI, what can it realistically achieve, and where are the limits?
- Enable broad access: Instead of expensive individual licenses, we rely on hosted Company GPTs. Everyone in the company, from intern to board member, can use the tool, learn, and discover their own use cases. This prevents shadow IT (using private ChatGPT accounts with company data!) and fosters a culture of innovation from the bottom up.
- Accompany use cases: Together with specialist departments, we identify pragmatic use cases that deliver quick wins – whether automating invoice analysis or optimizing document workflows.
2. The Unvarnished Truth: Your Biggest AI Blocker Is Your Legacy Systems
In the podcast, Florian and I talked a lot about the technical reality in SMEs. Companies are sitting on a treasure trove of data, but it is literally buried – in data silos, in legacy systems without interfaces, and in software whose manufacturers see AI more as a marketing stamp than a technical reality.
Before an AI can meaningfully access company data, the basics must be right:
- Data strategy: Where is which data? How is it structured?
- Interfaces (APIs): An AI needs “doors” to your systems. If these are missing, we as service providers have to build cumbersome “adapters” – tedious work that should actually be the job of software manufacturers.
- Enterprise Architecture: A clean interplay of systems prevents the “spaghetti code” that quickly returns through careless “low-coding.”
AI projects mercilessly expose the sins of the past. The good news: They finally create the necessary pressure to reduce these technical debts.
3. Humans Become AI Leaders
One of the biggest concerns is that AI will destroy jobs. I see it differently. The role of humans is not becoming superfluous, but more demanding. Routine tasks are eliminated, but strategic and qualitative responsibility grows.
The employee of the future becomes an “AI Leader.”
Their task is to guide the AI, critically question the results, and ensure quality. An experienced software architect is not replaced by AI coding tools – they become the reviewer who keeps an eye on the overall architecture and security. But a junior developer who just rubber-stamps AI suggestions will never acquire this competence. This presents us as a company with a completely new challenge in training the next generation.
4. What We Need Now: Courage, Pragmatism, and Homework
My appeal to companies, manufacturers, and politicians is clear:
- Companies: Have the courage to see AI as a holistic journey. Anchor the topic at the executive level and invest not only in tools but above all in empowering your employees and modernizing your IT landscape.
- Software manufacturers: Do your homework! Deliver clean, standardized interfaces (like MCP) so that the integration of AI systems becomes plug-and-play and not a tinkering session.
- Politics: Create clear, practical frameworks for data use (keyword: GDPR and anonymized data) and promote European champions like Mistral AI so that we don’t fall into complete dependence on US providers.
Conclusion
The conversation with Florian Mertl showed once again: Real progress with AI doesn’t come from the loudest hype, but from down-to-earth, strategic work. It’s about reducing complexity without trivializing. That is exactly our aspiration at innFactory and innFactory AI Consulting.
If you’re ready to look beyond the horizon and pragmatically and effectively anchor AI in your company, then listen to the podcast or contact us directly. We look forward to the exchange!