Modeling AI Applications

Enterprise AI applications are expressed through the following model artifacts on the C3 AI Platform:

1. Metadata – used to declare objects and their attributes
2. Annotations – used to indicate the underlying data store to persist an object’s data
3. Expressions – used to represent functions which perform mathematical operations on input data without programming logic
4. Programming Logic – used when necessary to express logic beyond metadata, annotations and expressions

These artifacts are used to represent the following interdependent “layers” of an application:

1. Data Exchange Object Model, Mappings, and Transformations – metadata representing a data model optimized/denormalized for data interchange between systems, mappings and data transformations to target an application object model.
2. Application Object Model – metadata representing a data model optimized for development of applications and machine learning models.
3. Methods – business logic representing actions (e.g., enroll a customer, maintain equipment) on objects expressed in code
4. Analytic Features – expressions which perform spatial and/or temporal mathematical operations on input data
5. AI Machine Learning Model – combination of analytic features and AI-ML algorithm (e.g., NLP pipeline, logistic regression, tensor flow…)
6. Security Services – annotations and expressions applied to Objects, Methods, Features and Machine Learning Models.
7. User Interface (Optional) – metadata representing user interface fields, their bindings to objects and object properties, and user interface objects (charts, grids, etc.,)

Model Extensibility

• Meta Data Extensions
• Code Extensions

Model Upgradeability

• Platform Upgrades
• Application Upgrades
• Preserving Customer Extensions

Common Platform for Application Developers and Data Scientists to Collaboratively Develop AI Applications

Data scientists typically work in isolation, developing and testing machine learning algorithms against small subsets of data provided by IT from one or more disparate source systems. The bulk of their time is spent on data cleansing and data normalization to represent the same entities, measures (units), states (e.g. status codes) and events consistently in time and across systems and to correlate (“join”) data across systems. The resulting algorithms, typically written in Python or R, do not conform to IT standards and require rewriting to different programming language such as a Java. Furthermore, the efficacy of the algorithm is sub-optimal since it has not been tuned against a representative production data set.

A platform is required to allow data scientists to develop, test, and tune algorithms in the programming language of their choice against a snapshot of all available production data. To accelerate development, data scientists can leverage work completed on the platform by data engineers and application developers to handle data cleansing, data normalization, object modeling and representation, microservices to focus on analytic feature development for classic machine learning or deep learning models.

The resulting machine learning algorithm is immediately deployable in production and available as a microservice through a standard RESTful API.

The C3 AI Platform is an example of a proven model-driven AI platform.

The C3 AI Platform allows small teams of five to ten application developers and data scientists to collaboratively develop, test and deploy large-scale production AI applications in one to three months. The platform is proven across thirty large scale deployments across industries including Energy, Manufacturing, Aerospace / Defense, Healthcare, and Financial Services. A representative large scale C3 AI Platform deployment processes AI inferences at a rate of a million messages per second against a petabyte-sized unified federated cloud data image aggregated from fifteen disparate corporate systems and a forty million sensor network. Global 1000 organizations have successfully used the platform to deploy full scale production deployments in 6 months and enterprise-wide digital transformations with over twenty AI applications in 24 to 36-month timeframes.