Skip to main content

Semantic Kernel Components

Semantic Kernel consists of modular components that work independently or together to power AI-driven workflows. This guide covers the primary components and how they interact within the kernel.

AI Service Connectors​

Provide an abstraction layer over various AI services (e.g., Chat Completion, Text Generation, Image/Audio conversions), so you can switch between providers seamlessly.

Functions and Plugins​

Plugins are containers of one or more functions, which:

  • Can be advertised to AI models for function calling.

  • Can be invoked inside prompt templates.

Prompt Templates​

These are flexible templates combining fixed instructions, user inputs, and plugin calls.

Their use cases include:

  • Kickstarting a Chat Completion interaction by rendering and invoking the model.

  • Function-like usage when registered as a plugin.

A typical Workflow might look like:

  1. Render the template

  2. Execute hardcoded plugin calls

  3. Run the Chat Completion model on the rendered result

  4. Return AI-generated result to the caller