top of page

Novel material analysis to reduce waste and increase efficiency

RefFIT delivers disruptive software for the material industry to efficiently analyze spectroscopy data.

Nicole Ruckstuhl, Alice De Baar, Willem Rischau, Iris Crassee

Could you explain what your startup does in a few sentences?

The ever-rising increase of production forms a large burden on the limited resources of our planet. Shortages in semiconductors and other raw materials are already causing disruption in supply chains. The solutions are firstly, to make more efficient use of materials to increase consistency and yield in production processes and produce higher quality products, and thereby produce less waste, secondly, to recycle materials better and more efficiently to reduce shortages. To reach these goals a profound understanding of materials is needed. RefFIT is innovative no-code software for the analysis of spectroscopic data used for material analysis. We help spectroscopy equipment manufacturers to implement better, faster and more complete analysis solutions. For the end users and clients in industry, using RefFIT means that the calibration and implementation phase is much shorter than with current solutions and they will be able to identify not just three or four predefined materials but a whole range. For example, in a sorting and recycling plant many more materials can be salvaged, while keeping a better view on possible pollutants and dangerous materials.

How is RefFIT different from its competitors?

Our competitors use mostly statistical and machine-learning approaches, which have to be calibrated using large amounts of data. This is a costly process taking many months and has to be repeated if the experiment evolves. We use scientific modeling, based on the laws of physics capturing the relevant parameters: the experimental, environmental and material parameters. This is an approach used in academic settings, but a complete software solution using this type of approach does not exist due to cost of implementation for every new situation. Therefore, we apply the best of both worlds: we use machine-learning principles to make the scientific models flexible and linkable. Hence, we can implement this type of scientific modeling faster and more easily making use of tested, linkable, scientific modules. We help our customers, both equipment manufacturers and end users, to identify a larger range of materials with better accuracy, make more responsible use of recycled and raw materials and deeply understand materials so that they can deliver high quality products to their clients with a better consistency and yield in the process.

Where do you see your startup in five years?

We are fully focused now on delivering our first solutions to key players in the spectroscopy market. Next, we will generalize our technology, moving away from spectroscopy towards a general data analysis solution. This solution still takes advantage of the scientific model-based approach powered with machine-learning principles, but will be suitable for all industrial engineering data analysis applications. The startup will have different branches by that time, with teams dedicated to the various applications and a large software team busy extending, testing and maintaining the ever-growing library of models.

What is the most useful advice you have received so far?

We were lucky that an excellent advisor joined the project at a very early stage. Our advisor instantly clicked both with us and with the project. He focused immediately on forming a diverse and dedicated team, which is one of the most important parts of a startup. Without this, it would not have worked for us. So, the best advice we got is to build a great team including advisors that you can trust and who trust you and who all have a strong belief in the project and work towards clear goals.


bottom of page