Microsoft Unifies .NET AI Agent Frameworks
Microsoft has launched a new Agent Framework for .NET developers, unifying its Semantic Kernel and AutoGen projects into a single SDK. The framework is designed for building enterprise-grade AI agents with capabilities for multi-agent orchestration, state management, and plugin integration. The move aims to simplify AI development for engineers in regulated industries like finance and insurance using the .NET stack.
- The unification combines Semantic Kernel's enterprise-grade features like type safety and state management with AutoGen's research-driven, multi-agent orchestration patterns. This was a direct response to developers wanting both the stability of Semantic Kernel and the innovative, experimental nature of AutoGen in a single SDK. - Before the merger, developers chose between Semantic Kernel for reliable, production-ready single agents integrated into existing enterprise applications, and AutoGen for exploring complex tasks with multiple, collaborating AI agents. The new framework is designed to handle both, from simple chatbots to sophisticated multi-agent systems. - The new unified framework introduces graph-based workflows, giving developers explicit control over how multiple agents execute tasks, including support for conditional routing and human-in-the-loop scenarios. It is built on `Microsoft.Extensions.AI` to standardize access to various model providers like OpenAI, Azure OpenAI, and local models through a common interface. - For engineers building on the side, the "Indie Hacker" community offers blueprints for building AI agents from scratch, often focusing on solving a specific personal pain point, like automating customer service calls. Key lessons from these builders often emphasize mimicking human reasoning in agent design and building in feedback loops for the agent to learn from its mistakes. - The NYC startup scene is seeing a surge in AI investment, accounting for 35% of all venture capital raised in the city. VCs like Thrive Capital are actively funding major AI players, while accelerators like Y Combinator have backed local AI agent startups such as Beacon Health, which builds AI agents to automate workflows for primary care providers. - In the broader AI agent market, venture capital funding is heavily concentrated on vertical SaaS solutions that solve high-value business problems. For instance, Sierra, a company building enterprise customer service agents founded by the former co-CEO of Salesforce, has reached a $10 billion valuation.
