The C3 AI Type System is a data object-centric abstraction layer that binds the various C3 AI Platform components, including infrastructure and services. It is both sufficient and necessary for developing and operating complex predictive analytics applications in the cloud.
The C3 AI Type System is the medium through which application developers and data scientists access the C3 AI Platform, C3 AI Virtual Data Lake, C3 AI Applications, and C3 AI Microservices. Examples of C3 AI Types include data objects (e.g., customer, product, supplier, contract, or sales opportunity) and their methods, application logic, and machine learning classifiers.
The C3 AI Type System allows programs, algorithms, and data structures – written in different programming languages, with different computational models, making different assumptions about the underlying infrastructure –to interoperate without knowledge of the underlying physical data models, data federation and storage models, interrelationships, dependencies, or the bindings between the various structural platform or cloud infrastructure services and components (e.g., RDBMS, No SQL, ETL, SPARK, Kafka, SQS, Kinesis, object models, classifiers, data science tools, etc.). The C3 AI Type System provides RESTful interfaces and programming language bindings to ALL underlying data and functionality.
Leveraging the C3 AI Type System, application developers and data scientists can focus on delivering immediate value, without the need to learn, integrate, or understand the complexities of the underlying systems. The C3 AI Type System enables programmers and data scientists to develop and deploy production AI, big data, and predictive analytics applications in one-tenth the time at one-tenth the cost of alternative technologies.
To improve manageability, Types support multiple object inheritance (allowing objects to inherit characteristics from one or more other objects). For example, a building might have characteristics of both a residential and commercial use building. The C3 AI Type system, through inherent dataflow capabilities, automatically triggers the appropriate processing of data changes by tracing implicit dependencies between objects, aggregates, analytic features and machine learning algorithms in a directed acyclic graph.
The C3 AI Type System supports combining AI and optimization algorithms to tackle challenging system simulation problems requiring dynamic probabilistic forecasts and constraint programming.
For example, C3 AI Inventory Optimization uses advanced AI and stochastic optimization techniques to account for both supply and demand-side uncertainties. C3 AI Inventory Optimization and the C3 AI Platform dynamically optimize inventory levels for individual parts on a near-real time basis to reduce excess inventory, maintain SLAs, and minimize the likelihood of stock-outs. This differs significantly from alternative approaches that account for stochasticity through static Monte Carlo methods, rather than the AI-based self-learning approach C3 AI Platform offers to better account for continuously changing circumstances.
Figure: Model-Driven Architecture Abstracts Underlying Platform Services through a simple Type Systems Interface.