Florian Quarre: We do have affinity to healthcare, finance, and retail primarily because we try to mix together the techno-functional aspects. On the functional side, quite a few of us either have past experience in healthcare, finance, or retail.
That allows us to understand the problem. It’s easy for us to relate to the problems and bring solutions we wish we had in the past or we know are going to be relevant to them in the future.
Sramana Mitra: Let’s do some use cases. Pick a few specific use cases that illustrate different aspects of your value proposition.
Florian Quarre: I will try to go in a way that it builds on top of the other. The first side of it is data organization at scale. A lot of our clients, when they scale, start having data that are siloed based on the line of business. You start representing similar type of information differently.
You could be a bank that has customers but maybe with different ways of talking about a potential person that you want to engage with. When you expand it, some of the organizations that we work with have hundreds of thousands of databases and other tables and data stores; it’s a massive ecosystem of data that lies out there. Yet that data is not integrated because they are not expressed the same way.
The first type of use case where we operate is looking into those data sources and from the definition of the metadata, be able to mesh it all together automatically in a way that we generate taxonomies and ontologies where you can start having a coherent source of truth on the definition of what a given business concept might be.
In this case, it’s customers. How the customer is represented across different lines of business or functions. As you start start using that data, you are able to navigate your data ecosystem fluidly and be in a position to ask questions.
First use case was the data organization at scale and data structuring. Second use case is information digitization. In many ways, it builds on top of the data organization. We see a lot of our clients deal with images. These are images for faxes or an Excel file in PDF. It could be invoices of technical documents.
The second use case that we do is use OCR and other techniques of NLP and information analysis to extract that data, render it electronically and then make it digital. It’s digital because it’s organized by reusing some of the taxonomies and ontologies that we would have worked on earlier. It defines a meaning for the information.
You can start looking into invoices, for example, that have a long list of details and relate it to purchase orders. In the healthcare space, that is as important. You can extract a tremendous volume of insights about patients within their medical records even though those records oftentimes are transmitted as an image.