Do you build agentic AI?

AI agents are autonomous entities powered by AI that can perform tasks, make decisions, and collaborate with other agents. Unlike traditional single-prompt LLM interactions, agents act as specialized workers with distinct roles, tools, and objectives.

Why build agentic systems?

• Automation - Handle complex, multi-step workflows autonomously
• Specialization - Assign agents to tasks they're optimized for (e.g., research, coding, analysis)
• Scalability - Deploy an "army" of agents to tackle large projects
• Collaboration - Enable agents to communicate and share insights

Leading agent frameworks comparison

When to choose each framework

When choosing a framework for building agents, you need to be aware of the tradeoff between customisability and ease of use.

For example, n8n is a no-code solution that is only really useful for automating simple workflows, while frameworks like LangGraph offer deeper control but with a steeper learning curve.

The closer you get to the raw LLM API requests, the more customisable you can make it – however, this will often come at the cost of additional development effort.

Microsoft Agent Framework ⭐️

Microsoft Agent Framework is Microsoft's open-source SDK for building agents and multi-agent workflows. It is the successor to Semantic Kernel and AutoGen, built by the same teams, and reached 1.0 with stable APIs and long-term support.

Microsoft Agent Framework shines when:

• You're building in .NET, or your team works across .NET and Python
• You need explicit, graph-based workflows where you control the path between agents, as well as open-ended agent collaboration
• You want to switch model providers (Microsoft Foundry, Azure OpenAI, OpenAI, Anthropic, Amazon Bedrock, Google Gemini, Ollama) without rewriting your agents
• Your agents need to call tools over MCP or talk to agents built with other frameworks over A2A
• You're deploying to Azure and want enterprise features like observability via OpenTelemetry

OpenAI Agents SDK

The OpenAI Agents SDK is ideal when:

• You require a straightforward, production-grade framework that's easy to learn
• Your existing stack already includes OpenAI models and you want seamless integration
• You need strong tracing and debugging capabilities built into the framework
• You're building for production environments

LangGraph

Consider LangGraph when:

• Your project involves intricate, recurring workflows requiring sophisticated state handling
• You're developing systems with multiple interconnected agents
• You have prior experience with the LangChain ecosystem
• You benefit from visualizing your agent workflows graphically
• You need enterprise-grade reliability

n8n

Choose n8n if:

• You need a no-code solution for building simple agent workflows
• You want a visual, drag-and-drop interface
• You have team members without coding experience who need to participate
• You're creating marketing, sales, or design team prototypes

Language limitations and considerations

Most of the leading agent frameworks today, including LangGraph and OpenAI Agents SDK, primarily use Python as their development language.

For C# developers, the best choice is Microsoft Agent Framework. It combines the enterprise-ready foundations of Semantic Kernel with the multi-agent orchestration of AutoGen, with first-class support for both .NET and Python.

Semantic Kernel and AutoGen are now in maintenance mode, so they only get bug fixes and security patches. Build new agents with Microsoft Agent Framework, and follow the official migration guides when you move existing projects.