Tacton AI journey

Important results from the project

The purpose of the project was to help Tacton get started with its AI effort. Models and infrastructure was developed in collaboration between Tacton and an expert partner and this project have led to Tacton gaining an increased understanding of AI and machine learning and the importance of having good data.

Expected long term effects

The project has led to models being developed for future use for increased business benefit as well as a knowledge lift in the company. The pilot models cannot currently be put into production as Tacton lacks sufficient data to train the models developed in the project. The models have instead been trained with realistic synthetic data, and can easily be adapted to actual data. The hope is that the models can be put into production when more data has been added, with only minor changes in model and code, and that these can be put on the same infrastructure base.

Approach and implementation

Through the project we have worked iteratively with analysis, development and evaluations with close reconciliations with project group and domain experts. Within the framework of the project, models were developed and infrastructure was set up to handle flows from ingestion of data to scalable ml model inference. We saw that a successful way of working was to lower the pace to enable opportunities to process and act on partial results, therefore the project was extended in calendar time.