The ModelOps solution should make it possible for companies to deploy a larger number of AI models. According to SAS, only 35 percent of analytics models currently reach deployment, and the rest are left behind in earlier stages of development.

SAS wants to combine its knowledge in the field of data science with AI, to deal with the “last mile challenges” that come with machine learning and AI in general. SAS ModelOps is delivered as a package, of which SAS Model Manager is a part. Support services that use open source and SAS models are also included. ModelOps Health Check Assessment is also part of the package; this is a solution for the optimal deployment of models.

ModelOps resembles the DevOps approach, and the holistic approach is carried over to ModelOps. This means that ModelOps comprises the entire life cycle of data analytics models. Where DevOps is all about a holistic approach to application development, ModelOps focuses on models instead of applications.

Solution for the entire process

In addition to the training and development process of AI models, SAS also wants to help companies with the preconditions necessary for the development of models. For example, there are functions for the backtracking of data to ensure that everything complies with the necessary regulations. There are also monitoring functions so that models can be continuously monitored for performance. If necessary, they can also be improved while they are already in use. 

With ModelOps Health Check Assessment, SAS offers suggestions for the optimal use of analytics models in the specific environments of users. To summarize, the ModelOps package comprises all requirements for companies to be able to develop, monitor and deploy data analytics models optimally.