About Semantic Kernel
Semantic Kernel is an open source development kit from Microsoft for building AI agents and adding models to existing code. It works in C#, Python and Java.
What it does
Semantic Kernel lets you describe your existing code to an AI model. When a request comes in, the model can choose the right function, and the kernel passes the result back. The kit also covers prompts, vector stores, text search and planning, along with hooks and filters for telemetry.
Key features
- Function calling over existing code
- Connectors for AI models and vector stores
- Prompt templates and plugins
- Hooks and filters for telemetry
- Agent and process frameworks
Who it's for
Enterprise teams and .NET developers who want AI features inside applications they already maintain. Java and Python developers who want a middleware layer, rather than a separate service, will also find it useful.
Pricing and good to know
The project is open source under the MIT licence. It is a library, so there is no subscription for the kit itself. The AI model provider you connect charges for usage. Microsoft's documentation says version 1.0 and later is committed to non-breaking changes across the three languages.
Highlights
- Let models call functions in your code
- Connect models and vector stores
- Reuse prompt templates and plugins
- Add telemetry with hooks and filters
- Build agents and multi-step processes
