Semantic Kernel
Microsoft's SDK for composing prompts, models, and tools into a plug-and-play architecture for AI apps.
Think of it like
A modular synthesizer where each knob is a skill, and you wire them together to make new instruments.
Example
Build a skill that translates text, chain it to a summarization skill, add tool use to fact-check sources. Swap in GPT or Claude, runs the same.
How it actually works
Semantic Kernel introduces 'skills' (prompt templates + tool definitions), 'planners' (ways to chain skills), and 'connectors' (to models and memory stores). It's orchestration-first: your app logic stays light, the framework handles the LLM calls. Strong on extensibility but can feel boilerplate-heavy.
For product teams
Reduces AI engineering boilerplate; lets you define skills once and reuse across agents.
For engineers
Plugin architecture with dependency injection; works with any LLM via adapters.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome