Microsoft has published a release candidate (RC) of its Semantic Kernel .NET SDK, the company’s open-source development kit for integrating large language models with conventional programming languages such as C#, Python, and Java.
The RC for the SDK was announced on December 5. As of December 7, Semantic Kernel was at the RC3 stage, with the interface getting close to its final v188.8.131.52 structure. With the latest updates, Microsoft set out to make the SDK simpler to use for both existing and new users. This involved renaming many classes and interfaces to better align with the rest of the industry and upgrading custom classes to existing .NET implementations, Microsoft said. Microsoft also has completely automated function calling with OpenAI, previously a multi-step process.
Microsoft said it has both increased the value of the kernel and made it easier to use by making it the “property bag” for the entire AI application. All of the elements of the application including multiple AI services, multiple plugins, and services such as loggers and HTTP handlers can be added to the kernel so that all components of Semantic Kernel can use them to perform AI requests.
Developers can use dependency injection in Semantic Kernel to create the kernel, and use a single YAML file to describe everything necessary for a prompt function, previously called semantic functions.