So the heart of our trust mechanism is personal relationships between people. We want to learn who trusts whom for recommendations for what kind of service and in which context. So to do that, we need to learn a lot about users, i.e. get quite a bit of data from them, and then build models that work in different domains.
That is a challenging task, because it basically requires a very general AI, which isn’t really available yet. So our approach is to start by modelling sector-by-sector, identifying common properties along the way and gradually generalising our models where possible, or otherwise choose algorithms and models depending on the use case.
One motivation of using a blockchain platform is to incentivise users to provide that data, but also to allow them to verifiably control access rights and conditions. So coming up with a model that would allow this, under a lot of constraints, was quite the task. We have a good foundation for this now.
So far, on the business side of things, apart from the taxi use case we started from, we’ve been onboarding several pilot clients, such as p2p and other lending services, but have a pipeline with clients also from very different sectors, like medical service platforms. Some of these, like the medical ones, also have quite strict requirements e.g. relating to privacy, which is why this aspect plays a large role in our model.
And now we’re targeting decentralised platforms and DApps, by providing our services via Oracles as well.