Semantic Kernel
The latest news from the Semantic Kernel team for developers
Latest posts
AI Agents for Beginners Course: 10 Lessons teaching you how to start building AI Agents
10 Lessons teaching everything you need to know to start building AI Agents Today we want to highlight the AI Agents For Beginners course that was released. 🔗https://github.com/microsoft/ai-agents-for-beginners/tree/main Start learning and building in the exciting world of AI Agents today! From the Semantic Kernel team, we look forward to seeing what you build!
Using OpenAI’s o3-mini Reasoning Model in Semantic Kernel
OpenAI’s o3-mini is a newly released small reasoning model (launched January 2025) that delivers advanced problem-solving capabilities at a fraction of the cost of previous models. It excels in STEM domains (science, math, coding) while maintaining low latency and cost similar to the earlier o1-mini model. This model is also available as Azure OpenAI Service, emphasizing its efficiency gains and new features like reasoning effort control and tool use. Throughout this post We'll explore how to use and other reasoning models with Semantic Kernel in both C# and Python. Key Features of OpenAI o3-mini: ...
Guest Blog: Step-by-Step Guide to Building a Portfolio Manager: A Multi-Agent System with Microsoft Semantic Kernel and Azure OpenAI
Today the Semantic Kernel team is excited to welcome back a guest author, Akshay Kokane to share his recent Medium article using Semantic Kernel and Azure OpenAI, showcasing a step-by-step guide to building a Portfolio Manager. We’ll turn it over to him to dive into his work below. In my previous blog, we went over how Semantic Kernel can be used to create a multi-agent system. Link. However, agent collaboration was really challenging, as we were not able to control how agents collaborated. We could set the termination strategy to decide when to stop collaboration between agents, but not how agents would pa...
Customer Case Study: How preezie’s AI shopping assistant is reshaping Blue Bungalow’s online store
Introduction Blue Bungalow, one of Australia’s leading fashion retailers, faced a common challenge in eCommerce—how to create a more engaging, seamless, and personalised shopping experience for customers online. They wanted to implement AI-powered assistance to provide personalised product recommendations, accurate sizing guidance, product comparisons, and instant answers to customer questions—replicating the ease and support of an in-store shopping experience. To bring this vision to life, preezie developed an AI shopping assistant, built on Semantic Kernel and Elasticsearch. https://www.youtube.com/watch?v=...
Guest Blog: Let your Copilot Declarative Agent think deep with DeepSeek-R1
Today we'd like to feature a guest author on our Semantic Kernel blog, Mahmoud Hassan, a Microsoft Valuable Professional (MVP) focused on AI. We'll turn it over to him to dive into his work below. In recent days, there has been significant attention in the AI community regarding DeepSeek-R1 and its capabilities. Many people are playing with it. For instance, Fabian Williams yesterday shared his experiment here: https://lnkd.in/dgZ8hjgB of running it locally. I thought, maybe today is my turn! However, I also thought it is an excellent opportunity to show a plugin design pattern I previously sha...
Using Azure OpenAI Chat Completion with data source and Function Calling
Azure OpenAI Chat Completion with data source provides powerful capabilities for integrating conversational AI into applications. However, using a data source and function calling in a single request is not supported yet. When both features are enabled, function calling is ignored, and only the data source is used. This presents a challenge when retrieving information, as a single request might not be sufficient to obtain an answer. This article shows how to address this limitation with custom retry mechanism. If a query remains unanswered, the system sequentially retries with different sources until the reque...
Semantic Kernel Roadmap H1 2025: Accelerating Agents, Processes, and Integration
As we move into the first half of 2025, I'm excited to share our ambitious roadmap that we hope will enable you to build even more sophisticated AI applications with Semantic Kernel. Agent Framework 1.0 By the end of Q1 2025, the SK Agent Framework will transition from preview to general availability (GA). This milestone marks a commitment to a stable, versioned API that minimizes breaking changes and supports production-grade applications. As highlighted in our recent community office hours meeting, our vision is to embrace an agent-first programming model. This approach has been shaped by extensive fe...
Using DeepSeek models in Microsoft Semantic Kernel
DeepSeek recently awed the AI community by open sourcing two new state-of-the-art models, the DeepSeek-V3 and a reasoning model, the DeepSeek-R1, that not only claim to be op-par with the most capable models from OpenAI but are also extremely cost-effective. We'd like to highlight the recent announcement from the Azure AI Foundry team highlighting DeepSeek R1 is now available on Azure AI Foundry and GitHub. DeepSeek is also available for use in Semantic Kernel, Azure AI Foundry and can be called via our inferencing connector as well. Semantic Kernel is thrilled to see such exciting development in...
Guest Blog: Creative Writing Assistant – a multi-agent app sample with Semantic Kernel & .NET Aspire
Today we’re featuring a guest author, Ricardo Niepel, who’s a Senior Technical Specialist at Microsoft. He’s written an article we’re sharing below; we’ll turn it over to Ricardo to share more! If “Creative Writing Assistant” sounds familiar, it might be because we've previously released a similar multi-agent application built in Python during our Ignite conference (Azure AI Foundry and the dev toolchain to infuse AI in all your apps | Ignite Session BRK119). Based on the invaluable feedback we received from all of you we are now excited to show a .NET version of this sample application. It illustrates how the...