Microsoft

Semantic Kernel

Enterprise SDK for LLM orchestration.

Semantic Kernel is Microsoft's polyglot SDK (.NET, Python, Java) for building AI agents that fit into enterprise codebases. It emphasizes plugins, planners, and integration with the broader Azure AI ecosystem.

Install

bash

pip install semantic-kernel

Quickstart

A minimal example to verify your setup.

python

import asyncio
from semantic_kernel import Kernel
from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion

async def main():
    kernel = Kernel()
    kernel.add_service(OpenAIChatCompletion(ai_model_id="gpt-4o-mini"))
    fn = kernel.add_function(
        plugin_name="writer",
        function_name="summarize",
        prompt="Summarize: {{$input}}",
    )
    result = await kernel.invoke(fn, input="Semantic Kernel is an SDK...")
    print(result)

asyncio.run(main())

Core concepts

Kernel

Central runtime that holds services, plugins, and memory. Every call goes through it for consistent telemetry.

Plugins

Native code and prompt functions packaged together so agents can call them as tools.

Planners

Automatically chain plugin calls to fulfill a goal — Handlebars, Function Calling, and Stepwise variants.

Agents framework

First-class agents with conversation threads, plus integration with Azure AI Agent Service.

Common use cases

›Adding AI to existing .NET / Java apps
›Enterprise agent platforms on Azure
›Plugin-based copilots over internal APIs
›Polyglot teams sharing one AI abstraction

Resources

← All frameworks Next: LangChain