Semantic Kernel (SK) is Microsoft enterprise framework for AI orchestration in .NET and Python. Key concepts: 1) Kernel: central orchestrator managing LLM calls, memory, and plugins. 2) Plugins: collections of functions (semantic functions = prompt templates, native functions = code). 3) Planners: AI-driven orchestration deciding which functions to call and in what order (Handlebars planner, Function Call planner). 4) Memory: semantic storage using embeddings for knowledge retrieval. Python example: kernel = Kernel(); kernel.add_service(AzureChatCompletion(service_id="chat", deployment_name="gpt-4o")); kernel.add_plugin(MathPlugin(), plugin_name="Math"); kernel.add_plugin(EmailPlugin(), plugin_name="Email"). Planner usage: planner = FunctionCallingStepwisePlanner(); result = await planner.invoke(kernel, "Calculate the total sales for Q1 and send a summary email to the finance team"). SK advantages over LangChain: tighter Azure integration, better C# support, more opinionated enterprise patterns. Suitable for: enterprises with .NET infrastructure, Azure deployments, complex multi-step workflows requiring reliable planning.