What drove the decision to develop an AI-based use case?
Even in the procurement of raw materials for production, more efficient companies have significantly more leeway than companies that make less profit from the raw materials they purchase. This shows that the competitive situation is already evident well before the supermarket shelf, and only those who are more efficient than their competitors in all value-adding processes will be able to pay higher prices for raw materials than the others for future production.
In order to achieve maximum profit, it is no longer enough to optimize KPIs in individual areas and apply naïve business formulas. Rather, the entire value chain must be considered if one wants to achieve more than just a local maximum profit. In order to achieve a global maximum profit, a large number of complex dependencies must be considered, which can no longer be controlled manually, even with a great deal of expertise.

What were your expectations or requirements for the AI application?
The customer expected an AI control center that would provide real-time strategic decisions for managing the optimum product mix in relation to all value-added processes and to the supply and demand situation in purchasing/sales.

Which partner did you rely on with which technology for the development?
The project was implemented by our Business Consulting, Business Intelligence, ERP and AI competence teams. For the individual business and data understanding at our client, we depended on the close exchange with their experts within the individual processes, corporate strategy and controlling. Two algorithms from the COIN-OR Foundation were adapted and further developed. Otherwise, the forecasting and optimization algorithms were developed in-house.

The use of AI solutions must also be accompanied by a change in competences. How have you dealt with this challenge in respect of your employees?
The role of planners and decision makers has changed in such a way that they now verify the recommendation given by the algorithm and weigh whether to modify it due to soft business strategies. A new "Enterprise Simulation" department has been created to support the AI control center.

In what timeframe did your company implement the AI solution?
The project was implemented in 3 phases within 18 months.

A conclusion to your AI solution:
The solution developed at our customer led to an increase in efficiency across the entire supply chain. The AI control center now controls all processes of routing and transport cost optimization to the individual plants, through the machine assignments of each preliminary product to the final product with client-specific assortment effects. This has resulted in our client posting a significant increase in profits without implementing any red pencil optimization.
Interesting side effects of our control station are production planning, routing, production monitoring and the ability to perform ad-hoc simulations for a wide variety of change scenarios within minutes.

One final piece of advice/tip to other entrepreneurs looking to apply AI:
From this project and our previous project experiences we would like to conclude 3 advices.

1. Since 80% of the AI project must be spent on business understanding, data understanding, data sourcing, there should always be a frequent exchange with the experts.
2. Stimulate idea generation via an initial workshop with experts in the company and with AI specialists, in order to then develop these further into target-oriented use cases. This also ensures that the technical and process-related integration into the company is clarified.
3. Never start an implementation of a use case as an elaborate concept paper, but rather according to the motto "Try Fast", "Fail Fast" and "Adapt Fast" always with a small delimited POC that proves the feasibility and effect of an implementation.

For more information, see our quick check with the workshop "AI Control Center"

More info at:
https://init-software.de/