Semantic Kernel / Planner

Semantic Kernel (SK) abstracts the complexity of working with LLMs behind a set of composable primitives. **Plugins** (formerly "skills") are collections of native code functions and prompt templates that the kernel can invoke. The **memory** system provides vector-backed context storage that plugins can read and write. The **planner** is the component that takes a high-level goal — "prepare a competitive analysis report on our top three rivals" — and automatically determines which plugins to call, in what order, to achieve it.

The planner is where Semantic Kernel differentiates from simpler orchestration libraries. Rather than requiring developers to hardcode execution sequences, the planner uses the LLM itself to generate a plan from the available plugins and then executes that plan step by step. This enables a class of enterprise application that can handle novel requests by recombining existing capabilities — a quality that makes SK particularly valuable in knowledge work automation, copilot development, and multi-step business process orchestration.

Semantic Kernel's primary enterprise context is the Microsoft AI ecosystem. It integrates natively with Azure OpenAI Service, Azure AI Foundry, and Microsoft Copilot Studio, and it has first-class support for .NET (C#) alongside Python and Java. Organizations building on M365, Dynamics 365, or Azure who want to extend those platforms with custom AI capabilities will find SK's tight ecosystem integration reduces development time significantly compared to ecosystem-agnostic frameworks like LangChain.