Weekly AI Roundup: Agent Frameworks, VS Code Skills, Azure Hosting
The AI section highlights updated agent frameworks, enhanced developer experiences for conversational AI applications, and new tools for integrating semantic data. Key releases include Microsoft’s latest Agent Framework, new agent customization options in VS Code, simplified ChatGPT hosting on Azure, and the Fabric IQ semantic data layer. These updates expand agent features and provide better support for context-aware solutions.
Microsoft Agent Framework: A Unified Approach to Enterprise AI Agents
Microsoft’s open-source Agent Framework is designed for building advanced AI agents and multi-agent systems, with support for both .NET and Python. The framework merges the stability and workflow features of Semantic Kernel with the modular design of AutoGen, making it straightforward to connect with LLMs, handle state, and integrate external tools. It focuses on maintainability, checkpointing, and human involvement, offering flexibility for long-running and complex agent operations in enterprise settings. Bringing together tools previously split across the ecosystem, Agent Framework streamlines agent development and operational support for both .NET and Python. Continuing themes from Azure AI Foundry and its ecosystem, this framework brings orchestration, hosting, and model routing under a single solution. It combines methods like Durable Task Extension with orchestration models to support Microsoft’s efforts to unify agent development for cloud and enterprise use.
Streamlining Agent Customization and Integration in Developer Tooling
Visual Studio Code now supports an Agent Skills feature, letting developers configure and manage agent abilities through markdown files directly in the editor. Skills can be switched on or off, providing flexible guidance to agents and simplifying context management. This streamlines the setup and tuning of agent capabilities in individual code projects. Recent updates to Agent Skills and the Skills.md standard complement Copilot and Azure AI Foundry improvements introduced last week. The current focus is on integrated skills editing in VS Code, building on positive community feedback about consistent, shareable configuration. These enhancements further the effort to make agent customization accessible across toolsets.
Hosting and Scaling ChatGPT Apps with Azure Functions
Developers can now deploy ChatGPT applications more easily using Azure Functions, the MCP server model (FastMCP for Python), and the Azure Developer CLI for setup and authentication. The step-by-step guide walks through backend service configuration, custom endpoint integration (like weather data), and embedding chat metadata. Azure’s serverless hosting means automated scaling, pay-by-usage pricing, and simple connection to the ChatGPT App Directory. Included examples and developer-mode tools offer clear paths for Python users building and testing advanced chat apps without dedicated infrastructure. Planned improvements include stronger authentication and increased support for MCP extensions on Azure. This matches recent trends around the Model Context Protocol and Azure AI Foundry’s model router, allowing developers to reliably link models, APIs, and backend resources for conversational agents. The walkthrough provides real code and approaches for running ChatGPT apps on Azure.
Enriching AI Agents with Semantic Data: Introducing Fabric IQ
Fabric IQ is a semantic modeling layer for Microsoft Fabric that helps agents understand business concepts and workflows. This abstraction layer maps raw data into entities and relationships, supports business logic with rules and validations, and enables monitoring through its ontology framework. The Fabric Graph allows for entity relationship mapping and supports both GQL and natural language queries, so agents can interact with business data more naturally. Full setup instructions and governance integration are included, making it easier for agents to incorporate organizational knowledge into their operation. Building on prior discussions about Fabric’s AI-powered workflows, this update introduces direct semantic modeling to the process, which supports better analytics and agent operations based on real organizational data and rules.