Select Packages Overview

Applications built on the C3 Agentic AI Platform combine packages from three sources: built-in platform capabilities, C3 AI plugins and applications, and external libraries through supported runtimes.

Understanding these package types helps you decide what to adopt, what to extend, and when to rely on external ecosystems.

Copy link to this section

Package types

There are three types of packages you can use in your C3 AI applications:

• Built-in packages: Core functionality that ships with the C3 Agentic AI Platform.
• Purpose-built packages: Pre-built C3 AI plugins and applications.
• External packages: Third-party libraries and custom runtimes.

Each package type serves a different purpose and provides different capabilities for your application development.

Copy link to this section

Built-in packages

Built-in packages provide the core functionality of the C3 Agentic AI Platform. These packages are automatically available in every application without requiring you to add any dependencies.

Built-in packages include capabilities for:

• UI development: Accelerate application development with the C3 AI UI Component Library, a curated library of out-of-the-box UI components including layout, forms, data visualization, and 3D components.
• Machine learning: Manage the complete ML lifecycle with model operations, distributed training, inference services, model evaluation, experiment tracking, and hyperparameter optimization.
• Data integration: Build data pipelines with ETL tools, data transformation frameworks, lineage tracking, and integration with external data sources.
• Observability: Monitor applications with performance tracking, logging analysis, job monitoring, and distributed tracing for debugging and optimization.
• GenAI: Integrate large language models from multiple providers, build AI agents with tool support, manage vector stores for embeddings, and process unstructured data.
• Workflow: Orchestrate processes with visual workflow design, automated deployment management, and complex workflow patterns for business logic.

You can use built-in package functionality immediately in your application. For example, the Persistable type is available without any configuration. For UI applications, you can use the uiComponentLibraryReact package to access pre-built UI components for building user interfaces.

To learn more about what functionality is available in built-in packages, see Packages Reference.

Copy link to this section

Purpose-built packages

Purpose-built packages are pre-built C3 AI packages designed for specific purposes. These include C3 AI Plugins and C3 AI Applications that you can inherit from to build your own solutions.

For example, you can inherit the reliability package to get the base functionality of the C3 AI Reliability application. This approach allows you to extend and customize existing C3 AI applications rather than building from scratch.

To learn how to discover and use available C3 AI Plugins and C3 AI Applications, see Application Compatibility and the catalogue of documented applications.

Copy link to this section

Third-party libraries and runtimes

Unlike C3 packages (built-in and purpose-built), you can also integrate third-party libraries from external ecosystems into your applications. You can add npm libraries for JavaScript development and Python libraries through custom runtimes for data science and machine learning workflows.

The C3 Agentic AI Platform uses a runtime system to manage these external dependencies. A runtime defines the execution context with specific library versions and dependencies. For JavaScript, the platform provides the js-webpack_c3 runtime for managing npm packages. For Python, you can create custom runtimes that specify the Python version, conda channels, and required packages.

Third-party libraries enable you to use the broader ecosystem of open-source and third-party tools within the C3 AI Type System while maintaining version consistency across development and production environments.

To learn more about working with external libraries and runtimes:

• For understanding how the runtime system works, see Understanding Action Engines.
• For creating custom Python runtimes, see Making a Runtime.
• For working with Python runtimes in JupyterLab, see Working with Runtimes.

Copy link to this section

Next steps

After understanding the types of packages available, you can:

• Review the built-in functionality to understand what the C3 Agentic AI Platform provides by default.
• Explore available C3 AI packages to see if existing solutions meet your needs.
• Add external packages when you need specific third-party functionality.