Rising skills shortages, increasing knowledge loss and complex production processes present major challenges for industrial companies. Individually developed AI assistance systems offer concrete solutions. They make experience-based knowledge usable, stabilize processes and open up new digital perspectives.
XITASO is a partner for high-end software engineering and develops tailored digital solutions for products, processes and platforms in industries such as mechanical engineering, medical technology and the public sector. With more than 240 employees at 13 locations, XITASO provides expertise in areas such as data science, AI, IT security and cloud engineering.
Skills shortages and knowledge loss as a challenge
Industrial companies are undergoing profound structural change. Several factors play a role here, including the demographic decline of qualified specialists, changing values in the working world and ongoing digitalization. The mechanical and plant engineering sector is particularly affected. Three parallel developments stand out:
- Training for demanding fields such as commissioning complex production systems takes months or even longer.
- Young professionals today have different expectations of their working environment. They are less likely to commit to a company for the long term. Business travel, for example to commission machinery at remote locations, is considered unattractive.
- The experience-based knowledge of departing employees is often only implicit and not documented. As a result, valuable information is lost.
Better understanding sources of error in complex processes
Industrial production processes are complex, so fluctuations in temperature or pressure can cause quality losses. One example: in plastics processing, a temperature error during extrusion can lead to several hundred meters of waste material.
The data underlying an error analysis is often incomplete, unstructured or difficult to interpret. Traditional methods reach their limits, and there is a lack of time and capacity to systematically analyze errors. As a result, production downtime is intensified or prolonged.
Challenges at a Glance
- Skills shortage in mechanical and plant engineering
- Difficult knowledge transfer during staff changes
- Production processes are complex and hard to control
- High effort required for data preparation and analysis
The solution: an AI assistance system to support employees
AI assistance systems are a contemporary solution to these challenges. Their goal is to support operational teams, detect errors at an early stage and take targeted countermeasures. Their unique strength lies in the combination of data-driven machine learning and the purposeful use of employees’ experience-based knowledge.
Tailored development instead of standard solutions
XITASO develops customized AI assistance systems as a tailored service. Close coordination between the departments involved is therefore essential during development in order to define all requirements at an early stage.
- First, XITASO’s developers identify potential use cases. Technical conditions and industry-specific requirements are taken into account, as well as existing data sources, system landscapes and organizational structures.
- They then evaluate the cases in terms of their economic potential, feasibility and expected benefits.
- Finally, the team designs the architecture of the solution. In doing so, it considers existing IoT platforms such as Azure, Bosch or CLX. If necessary, XITASO can implement on-premises solutions or hybrid models combining cloud and edge computing.
The concept is implemented iteratively (agile) in order to react flexibly to changing requirements and to make customer value visible at every stage of the project.
The technology of the AI assistance system
The AI assistance system combines three technological aspects: 1) anomaly detection, 2) prediction of relevant process parameters, and 3) the use of expert knowledge. Since faulty production data occurs only rarely due to already highly optimized processes, it is difficult to learn solely from historical data. For this reason, the development team incorporates the expertise of specialists. One approach is to ask operators directly about parameter adjustments they have made and their reasons for doing so, in order to derive generalizable rules.
These rules support the machine learning module in steering the learning process by guiding it toward insightful solutions. The method combines domain specific knowledge with data driven modeling. This ensures higher precision and robust performance even when data is insufficient. In this way, a hybrid solution emerges that delivers reliable decisions even when only limited data is available.
Knowledge storage and continuous improvement
The use of an AI assistance system reduces dependency on individual experienced employees, as their expertise becomes accessible through the assistance system. Companies can now utilize knowledge that previously only existed in people’s minds. In the long term, AI assistants create a comprehensive knowledge database that grows continuously and evolves steadily through new data and observations.
This database forms the basis for continuous process improvement, supports the training and onboarding of new employees, and enables deeper insights into the causes of process deviations. In this way, not only can known problems be addressed, but also new connections can be uncovered that would be difficult to detect using purely statistical methods.
AI assistance systems as a foundation for digital services and new business models
An AI assistance system provides the foundation for further digitization of the company through data driven maintenance processes, automated quality inspections, and new as a service business models. In particular, machine and plant manufacturers expand their traditional product portfolios with digital assistance functions.
The results are greater resilience and innovative strength. Companies that discover new connections and initiate product improvements based on collected data gain a clear advantage. Especially in a highly competitive and demanding industrial environment, these factors are key to future viability.
Summary of results
- Data driven production processes
- Less dependency on individuals
- Knowledge retention and documentation
- Future proof services for manufacturers
