Last week, the AI Practitioners Annual Conference by Äripäev took place, bringing together success stories from various industries, discussions on challenges, and practical advice on implementing AI development.
At the conference, Tammer’s IT Project Manager Silver Reinaas and IT Manager Indrek Kink presented our top AI innovation of the year. Last spring, in collaboration with Solita, our IT-team led by Silver implemented a machine learning model to predict product configuration errors early in the process, preventing costly issues down the line.
Previously, our process involved configuring the product, handing it over to production, identifying any non-conformities in manufacturing, and then analyzing those issues in a meeting afterwards. Now, we have an almost fully developed machine learning model that supports and significantly simplifies the entire process.
We have trained the model to identify non-conformities from the entered data. A crucial part of this effort has been validating the data used in the learning process, ensuring its quality and relevance. Fortunately, our Tammer IT team is thoroughly familiar with both the data and the software we use, which helps them understand the input data and recognize why potential errors occur.
Currently, the project is in the beta-testing phase with our production technologists, and we are soon entering the broader launch and integration phase so that the solution can support the work of our project department, production, and quality control.
The most effective AI adoption supports existing processes rather than reinventing new ones. There is no point in having AI as a separate goal for the company—there must be a real challenge that AI can solve.
