Semantic Kernel

Semantic Kernel is Microsoft's official SDK for integrating LLMs into enterprise applications, with first-class support for .NET/C#, Python, and Java. It provides a plugin architecture for combining AI reasoning with existing business logic.

Key use cases include:

• Building copilot-style experiences in enterprise .NET applications
• Integrating LLMs with existing business APIs and services through plugins
• Creating AI agents with planning capabilities that break complex tasks into steps
• Memory and knowledge management for context-aware AI assistants
• Enterprise applications using Azure OpenAI with tight Azure ecosystem integration
• Building AI features in Microsoft 365 and Teams extensions

Semantic Kernel is used by enterprise .NET teams building AI features into their existing applications. It is the foundation powering Microsoft 365 Copilot and is the recommended path for organizations in the Microsoft ecosystem.

The plugin architecture makes it easy to expose existing C# methods as tools that AI can invoke autonomously.

1. Install Semantic Kernel:

   # Python
   pip install semantic-kernel
   
   # .NET
   dotnet add package Microsoft.SemanticKernel

2. Configure with your AI service:

   from semantic_kernel import Kernel
   from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion
   
   kernel = Kernel()
   kernel.add_service(OpenAIChatCompletion(
       service_id='chat',
       ai_model_id='gpt-4o',
       api_key='sk-...'
   ))

3. Add plugins (functions the AI can call) and invoke:

   result = await kernel.invoke_prompt(
       'Tell me a joke about {{$topic}}',
       topic='cats'
   )

4. See Microsoft Learn — Semantic Kernel for full documentation.

Pricing: Semantic Kernel is free and open source (MIT). Works with OpenAI (pay-as-you-go) or Azure OpenAI (Azure subscription required). No additional licensing cost for the SDK itself.