CODE Magazine / Articles
Semantic Kernel
Articles filed under Semantic Kernel
Articles
Showing 1–5 of 5
-
The .NET AI Stack
Updated
Mike Yeager traces the evolution of .NET AI development—from early generative AI experiments through Semantic Kernel to the new Agent Framework (Microsoft.Agents.AI). With hands-on code examples, the article demonstrates how Microsoft.Extensions.AI and Agent Framework simplify agent integration, tool use, and workflows, making them the go-to stack for building flexible, future-proof AI applications in .NET.
-
Semantic Kernel Part 4: Agents
In "Semantic Kernel Part 4: Agents," Mike Yeager explores the use of agents within Semantic Kernel (SK) to tackle complex tasks by customizing Large Language Models (LLMs). Mike explains that agents function as specialized LLMs with specific capabilities, such as performing calculations or accessing tools like MATLAB, to produce more accurate and specialized outcomes. You'll learn about creating assistant agents for specific use cases, like tax calculation, and chat agents that collaborate to achieve tasks, exemplified through a software development scenario. These agents simplify processes by being modular and pre-configured, allowing developers to build more extensive, manageable systems while treating agents as source code.
-
Semantic Kernel Part 3: Advanced Topics
Mike details the evolution of his team's development of a Copilot system using Microsoft's Semantic Kernel (SK) framework for Large Language Models (LLMs). Initially, Mike describes their hands-on approach to building a Copilot capable of answering legal contract queries by manually injecting relevant data into prompts. However, as the project advanced, the team adopted GPT-4 and automatic function calling, significantly simplifying their code and enhancing functionality. Mike emphasizes the transformative impact of these new technologies, which allowed for more effective and efficient application development with minimal manual intervention.
-
Semantic Kernel 101: Part 2
Updated
Mike continues his tutorial on working with Semantic Kernel (SK), Microsoft's framework for working with Large Language Models (LLMs), including code examples since SK V1 has just released. Mike demonstrates how to create a deployment of a GPT-4 Large Language Model in the Azure portal. Learn to create a prompt in code and execute it against the model. Learn to execute your own C# code in the same way you executed the prompts, and how to chain multiple functions together in a pipeline.
-
Semantic Kernel 101
Mike Yeager introduces Semantic Kernel (SK), an open-source AI framework developed by Microsoft for .NET for working with large language models (LLMs) and specifically to help create Copilots. Yeager explains that SK serves three main purposes: to abstract the underlying LLMs, APIs, and tooling; to handle complex implementations in a generic way; and to facilitate the integration of user-generated content. He also discusses the benefits of using SK, such as its ability to create prompt templates, handle ad hoc scenarios, and assist with tasks like tokenization and text splitting. Despite still being in preview, Yeager suggests that SK is a powerful SDK that is worth exploring.