Evispot - an innovative startups, Phase 2
|Funding from Vinnova||SEK 900 000|
|Project duration||September 2018 - March 2019|
Purpose and goal
The project aimed to further develop the MVP developed in the Phase 1 project, to take Evispot one step closer our product vision - a platform specialized for AI credit decisions. During the prince hav we managed to get one step closer to this vision by further develop and implement this model live. In parallell have also gained feedback from customer regarding desired functionality and thereby has it been possible for us to understand what we should focus on product wise. The project aim has been fulfilled - which has been to further validate and confirm our product.
Expected results and effects
The project has been successful and we are now one step closer to our product vision. The MVP has been adapted for production and live-settings, we have developed structures for APIs, smooth updates and retraining of the models. During the project have we realized that the platform needs more parts than just the AI-models themselves, naming monitoring of model’s performance and user-interface as two examples of desired functions. The means, regarding the success of this project, we still have parts to develop before we have a fully scalable and scaleable platform.
Planned approach and implementation
The project has been performed in line with the established project plan. The first subproject primarily focused on getting our AI-models explainable. The second subproject included developing settings and APIs to enable our customer to easily get access to the models. We have also structures and settings enabling automatic retrainings of the model when ny data is available. We have not been able to perform an automatic retraining of the model, we are waiting for the enough turn-outs to initiate the retraining (it usually takes a few month before the loans are starting to being paid back).