AI services plattform to analyse agreements
Reference number | |
Coordinator | VIRTUAL INTELLIGENCE VQ AB |
Funding from Vinnova | SEK 1 908 000 |
Project duration | July 2018 - December 2019 |
Status | Completed |
Purpose and goal
The purpose of the project was to build a beta-version of an AI-platform that can analyse agreement documents. As we wanted to build services that could break down and analyse large agreements, as well as analyse several documents in parallell, we realised it was not sustainable to use existing cloud-based services. Instead, we saw that we needed to build our own platform where we could combine different AI-resources and its different tools in a powerful way. We have now succeeded with that, and we have a beta-version of this platform in place.
Expected results and effects
With this platform we can now see efficient ways of breaking down and analysing the different parts of an agreement in a very efficient way, with a system that minimises CPU usage and therefore the time required to run the AI models. It also means very efficient server usage which is very beneficial as we scale up the service.
Planned approach and implementation
We have had to analyse and compare several different AI-solutions, look for tolls for efficient annotations, and develop processes so that people with varying IT-level can assist in training the service. It has some times been difficult to find people with the right skills to aid in the work, as we’ve learnt that very few people have experience with this type of work. This has lead to extra work for us.