Semantic Kernel JS Concepts

The Semantic Kernel JS architecture is designed to be modular and extensible, allowing developers to build AI applications that can integrate various services and plugins seamlessly.

The architecture consists of several key components:

Diagram Components:

1) Your Application

A JavaScript application that utilizes AI capabilities.

It sits at the top layer and interacts with both the semantic-kernel and LLM service clients directly.

2) semantic-kernel

A core library that acts as an orchestration layer between your application and AI services.

Facilitates tasks such as prompt management, function chaining, and AI interaction logic.

3) LLM Clients and AI Services

Includes integrations for:

• OpenAI
• Azure AI Inference
• Ollama
• Other AI service providers

These services provide the underlying AI capabilities such as text generation, embeddings, and more.

4) @semantic-kernel/abstractions

A package containing common abstractions used by semantic-kernel to define and manage AI-related components.

5) @semantic-kernel/AI

A package responsible for interfacing directly with the AI service providers (OpenAI, Azure, Ollama, etc.).

Implements the concrete logic to communicate with external LLM APIs.

🔄 Flow Summary:

• Your application interacts with both semantic-kernel and the AI services.
• The semantic-kernel provides orchestration and abstraction.
• It communicates with the appropriate LLM clients via the @semantic-kernel/AI package.
• Underlying abstractions are handled through @semantic-kernel/abstractions.