VRP always donate a proportion of their working hours as part of their Pledge 1% initiative and chose the Noise Solution idea as the perfect project to work on a fully pro bono basis. We set to work evaluating potential natural language processing (NLP) models in terms of performance, accuracy and cloud resources. During the prototyping phase, VRP tested Salesforce Einstein Language and intelligent APIs from cloud vendors such as Microsoft Azure, Amazon Web Services (AWS) and Google Cloud. Tableau CRM was ultimately chosen to enable very high levels of accuracy, performance and security.
With the analytics platform and model decided, VRP next built a proof of concept (PoC) using the chosen model. The results proved that purely from the data, it could accurately determine whether teenagers needed easier or more difficult sessions. Finally, VRP deployed the model into production, connecting it to real data feeds ready for a period of live learning and refinement of the model.