Semantic Kernel
By Microsoft
Semantic Kernel is an open-source SDK from Microsoft for integrating large language models into applications, providing abstractions for plugins, memory, and planners that orchestrate multi-step AI tasks.
Definition
Semantic Kernel is an open-source SDK from Microsoft for integrating large language models into applications, providing abstractions for plugins, memory, and planners that orchestrate multi-step AI tasks.
Overview
Semantic Kernel lets developers combine natural-language prompts ("semantic functions") with native code functions into reusable plugins, which an LLM-driven planner can then orchestrate to accomplish a broader goal — similar in purpose to how LangChain and LangGraph let developers chain LLM calls and tools together, but built with Microsoft's own conventions and tight integration into the .NET and Azure ecosystems, as well as Python. It includes a memory abstraction for storing and retrieving context (often backed by a vector database), and a planner component that can dynamically decide which plugins to call in what order to satisfy a user's request, making it a foundation for building enterprise AI agents and copilots within Microsoft-centric technology stacks.
Key Features
- Plugin architecture combining prompts and native code functions
- Planner component for orchestrating multi-step tasks
- Memory abstraction backed by vector storage
- Support for both .NET and Python
- Tight integration with Azure AI services
Use Cases
Frequently Asked Questions
From the Blog
Semantic Search Explained: Beyond Keywords
Semantic search finds results by meaning rather than exact words, using vector embeddings so a query and a relevant document match even with no shared terms.
Read More AI & TechnologyWhat Is Semantic Search and How Does It Work?
Semantic search finds results by meaning, not keywords, using embeddings to represent text as vectors and matching queries to the closest ones in vector space.
Read More AI & TechnologyBLEU, ROUGE and Semantic Similarity: What Each Misses
BLEU rewards n-gram overlap with a reference, ROUGE rewards recall of reference content, and embedding similarity rewards being in the right semantic neighbourhood. Each can score a wrong answer highly, and knowing exactly how is what stops you gating a release on the wrong number.
Read More AI & TechnologyChunking Strategies for RAG: Fixed, Recursive and Semantic
Fixed-size chunking is the fastest baseline, recursive splitting respects document structure, and semantic chunking pays off only on unstructured prose. This compares the three on retrieval quality, shows where overlap earns its cost, and gives you an evaluation loop to decide on your own corpus.
Read More