METZEN AI expert Deborah Schwarze attended the 2bAHEAD Future Day in Leipzig to analyse key developments surrounding artificial intelligence, industry models and autonomous agent systems. The event highlighted how profoundly industrial value creation is set to change over the coming years.
Among the key topics discussed were Industry Models, brain computer interfaces and the growing use of specialised AI agents. A dense and highly inspiring day of insights.
What futures research can do – and what it cannot
Before evaluating the content of futures research, it is worth understanding the methodology behind it. In the context of 2bAHEAD, futures research is not about speculation, but about systematic probability assessment. Based on discussions with engineers, software developers and visionaries, development scenarios are formulated and evaluated according to their likelihood.
Much of what is presented on stages like this is technologically plausible. However, the speed and scale at which technologies are adopted in reality follow a different rhythm from developments in research laboratories or investor expectations. Failing to recognise that difference means confusing a roadmap with reality.
The reality gap is structural, not cyclical
Sven Gabor Janszky described a “reality gap”: rising insolvencies in traditional industries alongside massive growth in technology driven companies. At first glance, this may sound like a familiar narrative, but the analysis behind it is more precise than it appears.
The argument is not that the downturn is cyclical and will correct itself over time. The argument is that the gap is structural and continuing to accelerate. The difference is not created by isolated AI tools, but by the agentisation of entire processes. Decision making and execution logic are increasingly being transferred to autonomous AI systems.
Those who still treat this as a pilot project will only recognise the competitive shift once it has already happened.
Industry Models: the underestimated geopolitical factor
One development remained particularly significant beyond the event itself: Industry Models.
These are large, pre trained sector specific models for manufacturing, finance, medical services or e-government, built on deep industry data sets that go far beyond the capabilities of general models. Janszky referred to international developments already underway. Chinese organisations are actively developing and licensing such models for city and infrastructure management, while American platform providers are pursuing comparable approaches under different terminology.
The economic dimension is obvious. Companies that connect to an Industry Model at an early stage can achieve considerable productivity and efficiency gains.
Far less discussed is the geopolitical dimension. Whoever controls the model also controls its governance. That includes decisions about data access, decision logic and participation rights. This is precisely why the topic is not an abstract technology debate, but a strategic business decision.
The Klarna example as the most honest argument of the day
The most tangible statement of the day did not come from a future scenario, but from a real world example. Klarna rapidly automated large parts of its customer service with AI and discovered that efficiency gains and quality losses arrived hand in hand. Customer trust suffered noticeably.
The conclusion was clear: “AI right” beats “AI first”.
The same applies to the agent based approach. Janszky outlined scenarios involving ten to fifteen specialised agents working collaboratively. Technologically, the concept is convincing. In practice, however, it only works with clear governance, defined quality assurance loops and human oversight at the right points.
Speed without control is not an advantage. It is an operational risk.
Backcasting: convincing method, missing foundation
Backcasting was presented as a strategic methodology. The principle is to define a concrete vision for 2035 first and then work backwards to determine the actions required within the next twelve months.
Methodologically, the approach is convincing because it develops strategic logic from a target vision rather than from existing habits.
The practical challenge lies in the foundation. Many companies currently creating future scenarios still lack functioning AI governance, a clear data strategy and defined responsibilities for AI related decisions.
Even the best method requires robust structures. Without them, the future vision remains little more than a well designed presentation.
Robotics: the technology is coming, but more slowly
Robotics forecasts were also part of the Future Day discussions. The prediction of two humanoid robots per household within five to ten years appears technologically plausible when considering developments in motor skills, sensor systems and language models.
The actual adoption curve is likely to be slower. Not because the technology will fail to emerge, but because infrastructure, regulation, social acceptance and cost development all follow their own timelines.
The technological direction is clear. The timeframe remains far more complex.
What remains
The gap between companies structurally integrating AI and those still operating in isolated pilot projects is becoming a measurable competitive factor.
The answer is not the perfect strategy.
The answer is the first concrete step.
What this means for METZEN
Manufacturing is one of the core sectors for which Industry Models are being specifically developed. As a company with deep expertise in mechanical and plant engineering, METZEN operates directly within this environment.
That is why we actively engage with these developments. With a clear digital strategy and the ambition not to view Industry Models as an external trend, but as a strategic lever for our industries and customers.
The reality gap described at the event directly affects traditional industrial companies. Through the work of our AI expert, METZEN has already begun establishing the first structural foundations. Not as an isolated pilot project, but as a long term development with direct practical relevance.
That is exactly what makes the difference.
