This video demonstrates CognitiveScale’s Cortex AI Platform’s AI Personalization capabilities with Profile-of-One when building an AI application.
Profile of One is part of the Cortex AI engineering multi-modal, multi-persona platform that enables organizations to develop enterprise AI applications with observability and operationalization across the full life-cycle. Its visual programming and operating environment simplifies the development, deployment and management of transparent and trusted enterprise AI systems. Cortex connects data, orchestrates models, personalizes interactions, learns and infers, and ensures AI explainability and governance.
Profile-of-One plays a major role in driving Observable metrics on the business goals and KPIs and establishes value for the enterprise.
Profile-of-One begins by connecting data from a multitude of sources. We currently have a library of over 260 data connectors for various applications, including CRMs and RDBMSs. You can connect data from just about any of the disparate areas of your organization.
Once we have established connections we can define data sources and understand what data is available to the AI Engineer to construct an AI application.
Profile-of-One allows us to gather data from various sources and connect them. It is a way of establishing a visualization of any kind of entity using different attributes and grouping the attributes together to make them accessible to developers.
But it isn’t just a data aggregation method. It’s actually a beautiful way to create a temporal-based representation of any type of entity. Entities come in many forms. They can be customers or claims, for instance.
Three Classes of Information
The profile contains three classes of information that we are gathering from wherever we can throughout the organization.
One class of information is Declared attributes. This contains values such as marriage status, age, height, etc. These are things we know about the individual. This data comes from places like a CRM or some other data management system.
We can also pull in other data such as Observed Behavior. Observed Behavior includes things like when was the last time the customer visited the website and what actions did they take. Now, we have knowledge of what they do and we can observe their characteristics and we can make decisions based upon their actions.
When we take Declared and Observed information we can then develop Insights from those values using analytic models and other techniques. These Insights can be stored in the profile and this is when the profile becomes a temporal representation of the information we have gathered.
The profile then becomes a very powerful place where we can take things like Risk models or marketing ideas—for instance, if someone is searching for a mortgage—and scan that data and run analytics so it can be used to drive a hyper-personalized interaction with the individual.
Profile-of-One allows you to build an application using Declared, Observed and Inferred data from multiple data points. And the system continuously learns to get the most out of your interactions at an individual level.