Building Interactive Agent UIs with AG-UI and Microsoft Agent Framework
Introduction Picture this: You've built an AI agent that analyzes financial data. A user uploads a quarterly report and asks: "What are the top three expense categories?" Behind the scenes, your agent parses the spreadsheet, aggregates thousands of rows, and generates visualizations. All in 20 seconds. But the user? They see a loading spinner. Nothing else. No "reading file" message, no "analyzing data" indicator, no hint that progress is being made. They start wondering: Is it frozen? Should I refresh? The problem isn't the agent's capabilities - it's the communication gap between the agent running on the backend and the user interface. When agents perform multi-step reasoning, call external APIs, or execute complex tool chains, users deserve to see what's happening. They need streaming updates, intermediate results, and transparent progress indicators. Yet most agent frameworks force developers to choose between simple request/response patterns or building custom solutions to stream updates to their UIs. This is where AG-UI comes in. AG-UI is a fairly new event-based protocol that standardizes how agents communicate with user interfaces. Instead of every framework and development team inventing their own streaming solution, AG-UI provides a shared vocabulary of structured events that work consistently across different agent implementations. When an agent starts processing, calls a tool, generates text, or encounters an error, the UI receives explicit, typed events in real time. The beauty of AG-UI is its framework-agnostic design. While this blog post demonstrates integration with Microsoft Agent Framework (MAF), the same AG-UI protocol works with LangGraph, CrewAI, or any other compliant framework. Write your UI code once, and it works with any AG-UI-compliant backend. (Note: MAF supports both Python and .NET - this blog post focuses on the Python implementation.) TL;DR The Problem: Users don't get real-time updates while AI agents work behind the scenes - no progress indicators, no transparency into tool calls, and no insight into what's happening. The Solution: AG-UI is an open, event-based protocol that standardizes real-time communication between AI agents and user interfaces. Instead of each development team and framework inventing custom streaming solutions, AG-UI provides a shared vocabulary of structured events (like TOOL_CALL_START, TEXT_MESSAGE_CONTENT, RUN_FINISHED) that work across any compliant framework. Key Benefits: Framework-agnostic - Write UI code once, works with LangGraph, Microsoft Agent Framework, CrewAI, and more Real-time observability - See exactly what your agent is doing as it happens Server-Sent Events - Built on standard HTTP for universal compatibility Protocol-managed state - No manual conversation history tracking In This Post: You'll learn why AG-UI exists, how it works, and build a complete working application using Microsoft Agent Framework with Python - from server setup to client implementation. What You'll Learn This blog post walks through: Why AG-UI exists - how agent-UI communication has evolved and what problems current approaches couldn't solve How the protocol works - the key design choices that make AG-UI simple, reliable, and framework-agnostic Protocol architecture - the generic components and how AG-UI integrates with agent frameworks Building an AG-UI application - a complete working example using Microsoft Agent Framework with server, client, and step-by-step setup Understanding events - what happens under the hood when your agent runs and how to observe it Thinking in events - how building with AG-UI differs from traditional APIs, and what benefits this brings Making the right choice - when AG-UI is the right fit for your project and when alternatives might be better Estimated reading time: 15 minutes Who this is for: Developers building AI agents who want to provide real-time feedback to users, and teams evaluating standardized approaches to agent-UI communication To appreciate why AG-UI matters, we need to understand the journey that led to its creation. Let's trace how agent-UI communication has evolved through three distinct phases. The Evolution of Agent-UI Communication AI agents have become more capable over time. As they evolved, the way they communicated with user interfaces had to evolve as well. Here's how this evolution unfolded. Phase 1: Simple Request/Response In the early days of AI agent development, the interaction model was straightforward: send a question, wait for an answer, display the result. This synchronous approach mirrored traditional API calls and worked fine for simple scenarios. # Simple, but limiting response = agent.run("What's the weather in Paris?") display(response) # User waits... and waits... Works for: Quick queries that complete in seconds, simple Q&A interactions where immediate feedback and interactivity aren't critical. Breaks down: When agents need to call multiple tools, perform multi-step reasoning, or process complex queries that take 30+ seconds. Users see nothing but a loading spinner, with no insight into what's happening or whether the agent is making progress. This creates a poor user experience and makes it impossible to show intermediate results or allow user intervention. Recognizing these limitations, development teams began experimenting with more sophisticated approaches. Phase 2: Custom Streaming Solutions As agents became more sophisticated, teams recognized the need for incremental feedback and interactivity. Rather than waiting for the complete response, they implemented custom streaming solutions to show partial results as they became available. # Every team invents their own format for chunk in agent.stream("What's the weather?"): display(chunk) # But what about tool calls? Errors? Progress? This was a step forward for building interactive agent UIs, but each team solved the problem differently. Also, different frameworks had incompatible approaches - some streamed only text tokens, others sent structured JSON, and most provided no visibility into critical events like tool calls or errors. The problem: No standardization across frameworks - client code that works with LangGraph won't work with Crew AI, requiring separate implementations for each agent backend Each implementation handles tool calls differently - some send nothing during tool execution, others send unstructured messages Complex state management - clients must track conversation history, manage reconnections, and handle edge cases manually The industry needed a better solution - a common protocol that could work across all frameworks while maintaining the benefits of streaming. Phase 3: Standardized Protocol (AG-UI) AG-UI emerged as a response to the fragmentation problem. Instead of each framework and development team inventing their own streaming solution, AG-UI provides a shared vocabulary of events that work consistently across different agent implementations. # Standardized events everyone understands async for event in agent.run_stream("What's the weather?"): if event.type == "TEXT_MESSAGE_CONTENT": display_text(event.delta) elif event.type == "TOOL_CALL_START": show_tool_indicator(event.tool_name) elif event.type == "TOOL_CALL_RESULT": show_tool_result(event.result) The key difference is structured observability. Rather than guessing what the agent is doing from unstructured text, clients receive explicit events for every stage of execution: when the agent starts, when it generates text, when it calls a tool, when that tool completes, and when the entire run finishes. What's different: A standardized vocabulary of event types, complete observability into agent execution, and framework-agnostic clients that work with any AG-UI-compliant backend. You write your UI code once, and it works whether the backend uses Microsoft Agent Framework, LangGraph, or any other framework that speaks AG-UI. Now that we've seen why AG-UI emerged and what problems it solves, let's examine the specific design decisions that make the protocol work. These choices weren't arbitrary - each one addresses concrete challenges in building reliable, observable agent-UI communication. The Design Decisions Behind AG-UI Why Server-Sent Events (SSE)? Aspect WebSockets SSE (AG-UI) Complexity Bidirectional Unidirectional (simpler) Firewall/Proxy Sometimes blocked Standard HTTP Reconnection Manual implementation Built-in browser support Use case Real-time games, chat Agent responses (one-way) For agent interactions, you typically only need server→client communication, making SSE a simpler choice. SSE solves the transport problem - how events travel from server to client. But once connected, how does the protocol handle conversation state across multiple interactions? Why Protocol-Managed Threads? # Without protocol threads (client manages): conversation_history = [] conversation_history.append({"role": "user", "content": message}) response = agent.complete(conversation_history) conversation_history.append({"role": "assistant", "content": response}) # Complex, error-prone, doesn't work with multiple clients # With AG-UI (protocol manages): thread = agent.get_new_thread() # Server creates and manages thread agent.run_stream(message, thread=thread) # Server maintains context # Simple, reliable, shareable across clients With transport and state management handled, the final piece is the actual messages flowing through the connection. What information should the protocol communicate, and how should it be structured? Why Standardized Event Types? Instead of parsing unstructured text, clients get typed events: RUN_STARTED - Agent begins (start loading UI) TEXT_MESSAGE_CONTENT - Text chunk (stream to user) TOOL_CALL_START - Tool invoked (show "searching...", "calculating...") TOOL_CALL_RESULT - Tool finished (show result, update UI) RUN_FINISHED - Complete (hide loading) This lets UIs react intelligently without custom parsing logic. Now that we understand the protocol's design choices, let's see how these pieces fit together in a complete system. Architecture Overview Here's how the components interact: The communication between these layers relies on a well-defined set of event types. Here are the core events that flow through the SSE connection: Core Event Types AG-UI provides a standardized set of event types to describe what's happening during an agent's execution: RUN_STARTED - agent begins execution TEXT_MESSAGE_START, TEXT_MESSAGE_CONTENT, TEXT_MESSAGE_END - streaming segments of text TOOL_CALL_START, TOOL_CALL_ARGS, TOOL_CALL_END, TOOL_CALL_RESULT - tool execution events RUN_FINISHED - agent has finished execution RUN_ERROR - error information This model lets the UI update as the agent runs, rather than waiting for the final response. The generic architecture above applies to any AG-UI implementation. Now let's see how this translates to Microsoft Agent Framework. AG-UI with Microsoft Agent Framework While AG-UI is framework-agnostic, this blog post demonstrates integration with Microsoft Agent Framework (MAF) using Python. MAF is available in both Python and .NET, giving you flexibility to build AG-UI applications in your preferred language. Understanding how MAF implements the protocol will help you build your own applications or work with other compliant frameworks. Integration Architecture The Microsoft Agent Framework integration involves several specialized layers that handle protocol translation and execution orchestration: Understanding each layer: FastAPI Endpoint - Handles HTTP requests and establishes SSE connections for streaming AgentFrameworkAgent - Protocol wrapper that translates between AG-UI events and Agent Framework operations Orchestrators - Manage execution flow, coordinate tool calling sequences, and handle state transitions ChatAgent - Your agent implementation with instructions, tools, and business logic ChatClient - Interface to the underlying language model (Azure OpenAI, OpenAI, or other providers) The good news? When you call add_agent_framework_fastapi_endpoint, all the middleware layers are configured automatically. You simply provide your ChatAgent, and the integration handles protocol translation, event streaming, and state management behind the scenes. Now that we understand both the protocol architecture and the Microsoft Agent Framework integration, let's build a working application. Hands-On: Building Your First AG-UI Application This section demonstrates how to build an AG-UI server and client using Microsoft Agent Framework and FastAPI. Prerequisites Before building your first AG-UI application, ensure you have: Python 3.10 or later installed Basic understanding of async/await patterns in Python Azure CLI installed and authenticated (az login) Azure OpenAI service endpoint and deployment configured (setup guide) Cognitive Services OpenAI Contributor role for your Azure OpenAI resource You'll also need to install the AG-UI integration package: pip install agent-framework-ag-ui --pre This automatically installs agent-framework-core, fastapi, and uvicorn as dependencies. With your environment configured, let's create the server that will host your agent and expose it via the AG-UI protocol. Building the Server Let's create a FastAPI server that hosts an AI agent and exposes it via AG-UI: # server.py import os from typing import Annotated from dotenv import load_dotenv from fastapi import FastAPI from pydantic import Field from agent_framework import ChatAgent, ai_function from agent_framework.azure import AzureOpenAIChatClient from agent_framework_ag_ui import add_agent_framework_fastapi_endpoint from azure.identity import DefaultAzureCredential # Load environment variables from .env file load_dotenv() # Validate environment configuration openai_endpoint = os.getenv("AZURE_OPENAI_ENDPOINT") model_deployment = os.getenv("AZURE_OPENAI_DEPLOYMENT_NAME") if not openai_endpoint: raise RuntimeError("Missing required environment variable: AZURE_OPENAI_ENDPOINT") if not model_deployment: raise RuntimeError("Missing required environment variable: AZURE_OPENAI_DEPLOYMENT_NAME") # Define tools the agent can use @ai_function def get_order_status( order_id: Annotated[str, Field(description="The order ID to look up (e.g., ORD-001)")] ) -> dict: """Look up the status of a customer order. Returns order status, tracking number, and estimated delivery date. """ # Simulated order lookup orders = { "ORD-001": {"status": "shipped", "tracking": "1Z999AA1", "eta": "Jan 25, 2026"}, "ORD-002": {"status": "processing", "tracking": None, "eta": "Jan 23, 2026"}, "ORD-003": {"status": "delivered", "tracking": "1Z999AA3", "eta": "Delivered Jan 20"}, } return orders.get(order_id, {"status": "not_found", "message": "Order not found"}) # Initialize Azure OpenAI client chat_client = AzureOpenAIChatClient( credential=DefaultAzureCredential(), endpoint=openai_endpoint, deployment_name=model_deployment, ) # Configure the agent with custom instructions and tools agent = ChatAgent( name="CustomerSupportAgent", instructions="""You are a helpful customer support assistant. You have access to a get_order_status tool that can look up order information. IMPORTANT: When a user mentions an order ID (like ORD-001, ORD-002, etc.), you MUST call the get_order_status tool to retrieve the actual order details. Do NOT make up or guess order information. After calling get_order_status, provide the actual results to the user in a friendly format.""", chat_client=chat_client, tools=[get_order_status], ) # Initialize FastAPI application app = FastAPI( title="AG-UI Customer Support Server", description="Interactive AI agent server using AG-UI protocol with tool calling" ) # Mount the AG-UI endpoint add_agent_framework_fastapi_endpoint(app, agent, path="/chat") def main(): """Entry point for the AG-UI server.""" import uvicorn print("Starting AG-UI server on http://localhost:8000") uvicorn.run(app, host="0.0.0.0", port=8000, log_level="info") # Run the application if __name__ == "__main__": main() What's happening here: We define a tool: get_order_status with the AI_function decorator Use Annotated and Field for parameter descriptions to help the agent understand when and how to use the tool We create an Azure OpenAI chat client with credential authentication The ChatAgent is configured with domain-specific instructions and the tools parameter add_agent_framework_fastapi_endpoint automatically handles SSE streaming and tool execution The server exposes the agent at the /chat endpoint Note: This example uses Azure OpenAI, but AG-UI works with any chat model. You can also integrate with Azure AI Foundry's model catalog or use other LLM providers. Tool calling is supported by most modern LLMs including GPT-4, GPT-4o, and Claude models. To run this server: # Set your Azure OpenAI credentials export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com/" export AZURE_OPENAI_DEPLOYMENT_NAME="gpt-4o" # Start the server python server.py With your server running and exposing the AG-UI endpoint, the next step is building a client that can connect and consume the event stream. Streaming Results to Clients With the server running, clients can connect and stream events as the agent processes requests. Here's a Python client that demonstrates the streaming capabilities: # client.py import asyncio import os from dotenv import load_dotenv from agent_framework import ChatAgent, FunctionCallContent, FunctionResultContent from agent_framework_ag_ui import AGUIChatClient # Load environment variables from .env file load_dotenv() async def interactive_chat(): """Interactive chat session with streaming responses.""" # Connect to the AG-UI server base_url = os.getenv("AGUI_SERVER_URL", "http://localhost:8000/chat") print(f"Connecting to: {base_url}\n") # Initialize the AG-UI client client = AGUIChatClient(endpoint=base_url) # Create a local agent representation agent = ChatAgent(chat_client=client) # Start a new conversation thread conversation_thread = agent.get_new_thread() print("Chat started! Type 'exit' or 'quit' to end the session.\n") try: while True: # Collect user input user_message = input("You: ") # Handle empty input if not user_message.strip(): print("Please enter a message.\n") continue # Check for exit commands if user_message.lower() in ["exit", "quit", "bye"]: print("\nGoodbye!") break # Stream the agent's response print("Agent: ", end="", flush=True) # Track tool calls to avoid duplicate prints seen_tools = set() async for update in agent.run_stream(user_message, thread=conversation_thread): # Display text content if update.text: print(update.text, end="", flush=True) # Display tool calls and results for content in update.contents: if isinstance(content, FunctionCallContent): # Only print each tool call once if content.call_id not in seen_tools: seen_tools.add(content.call_id) print(f"\n[Calling tool: {content.name}]", flush=True) elif isinstance(content, FunctionResultContent): # Only print each result once result_id = f"result_{content.call_id}" if result_id not in seen_tools: seen_tools.add(result_id) result_text = content.result if isinstance(content.result, str) else str(content.result) print(f"[Tool result: {result_text}]", flush=True) print("\n") # New line after response completes except KeyboardInterrupt: print("\n\nChat interrupted by user.") except ConnectionError as e: print(f"\nConnection error: {e}") print("Make sure the server is running.") except Exception as e: print(f"\nUnexpected error: {e}") def main(): """Entry point for the AG-UI client.""" asyncio.run(interactive_chat()) if __name__ == "__main__": main() Key features: The client connects to the AG-UI endpoint using AGUIChatClient with the endpoint parameter run_stream() yields updates containing text and content as they arrive Tool calls are detected using FunctionCallContent and displayed with [Calling tool: ...] Tool results are detected using FunctionResultContent and displayed with [Tool result: ...] Deduplication logic (seen_tools set) prevents printing the same tool call multiple times as it streams Thread management maintains conversation context across messages Graceful error handling for connection issues To use the client: # Optional: specify custom server URL export AGUI_SERVER_URL="http://localhost:8000/chat" # Start the interactive chat python client.py Example Session: Connecting to: http://localhost:8000/chat Chat started! Type 'exit' or 'quit' to end the session. You: What's the status of order ORD-001? Agent: [Calling tool: get_order_status] [Tool result: {"status": "shipped", "tracking": "1Z999AA1", "eta": "Jan 25, 2026"}] Your order ORD-001 has been shipped! - Tracking Number: 1Z999AA1 - Estimated Delivery Date: January 25, 2026 You can use the tracking number to monitor the delivery progress. You: Can you check ORD-002? Agent: [Calling tool: get_order_status] [Tool result: {"status": "processing", "tracking": null, "eta": "Jan 23, 2026"}] Your order ORD-002 is currently being processed. - Status: Processing - Estimated Delivery: January 23, 2026 Your order should ship soon, and you'll receive a tracking number once it's on the way. You: exit Goodbye! The client we just built handles events at a high level, abstracting away the details. But what's actually flowing through that SSE connection? Let's peek under the hood. Event Types You'll See As the server streams back responses, clients receive a series of structured events. If you were to observe the raw SSE stream (e.g., using curl), you'd see events like: curl -N http://localhost:8000/chat \ -H "Content-Type: application/json" \ -H "Accept: text/event-stream" \ -d '{"messages": [{"role": "user", "content": "What'\''s the status of order ORD-001?"}]}' Sample event stream (with tool calling): data: {"type":"RUN_STARTED","threadId":"eb4d9850-14ef-446c-af4b-23037acda9e8","runId":"chatcmpl-xyz"} data: {"type":"TEXT_MESSAGE_START","messageId":"e8648880-a9ff-4178-a17d-4a6d3ec3d39c","role":"assistant"} data: {"type":"TOOL_CALL_START","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","toolCallName":"get_order_status","parentMessageId":"e8648880-a9ff-4178-a17d-4a6d3ec3d39c"} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"{\""} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"order"} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"_id"} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"\":\""} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"ORD"} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"-"} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"001"} data: {"type":"TOOL_CALL_ARGS","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","delta":"\"}"} data: {"type":"TOOL_CALL_END","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y"} data: {"type":"TOOL_CALL_RESULT","messageId":"f048cb0a-a049-4a51-9403-a05e4820438a","toolCallId":"call_GTWj2N3ZyYiiQIjg3fwmiQ8y","content":"{\"status\": \"shipped\", \"tracking\": \"1Z999AA1\", \"eta\": \"Jan 25, 2026\"}","role":"tool"} data: {"type":"TEXT_MESSAGE_START","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","role":"assistant"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":"Your"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":" order"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":" ORD"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":"-"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":"001"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":" has"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":" been"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":" shipped"} data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf","delta":"!"} ... (additional TEXT_MESSAGE_CONTENT events streaming the response) ... data: {"type":"TEXT_MESSAGE_END","messageId":"8215fc88-8cb6-4ce4-8bdb-a8715dcd26cf"} data: {"type":"RUN_FINISHED","threadId":"eb4d9850-14ef-446c-af4b-23037acda9e8","runId":"chatcmpl-xyz"} Understanding the flow: RUN_STARTED - Agent begins processing the request TEXT_MESSAGE_START - First message starts (will contain tool calls) TOOL_CALL_START - Agent invokes the get_order_status tool Multiple TOOL_CALL_ARGS events - Arguments stream incrementally as JSON chunks ({"order_id":"ORD-001"}) TOOL_CALL_END - Tool invocation structure complete TOOL_CALL_RESULT - Tool execution finished with result data TEXT_MESSAGE_START - Second message starts (the final response) Multiple TEXT_MESSAGE_CONTENT events - Response text streams word-by-word TEXT_MESSAGE_END - Response message complete RUN_FINISHED - Entire run completed successfully This granular event model enables rich UI experiences - showing tool execution indicators ("Searching...", "Calculating..."), displaying intermediate results, and providing complete transparency into the agent's reasoning process. Seeing the raw events helps, but truly working with AG-UI requires a shift in how you think about agent interactions. Let's explore this conceptual change. The Mental Model Shift Traditional API Thinking # Imperative: Call and wait response = agent.run("What's 2+2?") print(response) # "The answer is 4" Mental model: Function call with return value AG-UI Thinking # Reactive: Subscribe to events async for event in agent.run_stream("What's 2+2?"): match event.type: case "RUN_STARTED": show_loading() case "TEXT_MESSAGE_CONTENT": display_chunk(event.delta) case "RUN_FINISHED": hide_loading() Mental model: Observable stream of events This shift feels similar to: Moving from synchronous to async code Moving from REST to event-driven architecture Moving from polling to pub/sub This mental shift isn't just philosophical - it unlocks concrete benefits that weren't possible with request/response patterns. What You Gain Observability # You can SEE what the agent is doing TOOL_CALL_START: "get_order_status" TOOL_CALL_ARGS: {"order_id": "ORD-001"} TOOL_CALL_RESULT: {"status": "shipped", "tracking": "1Z999AA1", "eta": "Jan 25, 2026"} TEXT_MESSAGE_START: "Your order ORD-001 has been shipped..." Interruptibility # Future: Cancel long-running operations async for event in agent.run_stream(query): if user_clicked_cancel: await agent.cancel(thread_id, run_id) break Transparency # Users see the reasoning process "Looking up order ORD-001..." "Order found: Status is 'shipped'" "Retrieving tracking information..." "Your order has been shipped with tracking number 1Z999AA1..." To put these benefits in context, here's how AG-UI compares to traditional approaches across key dimensions: AG-UI vs. Traditional Approaches Aspect Traditional REST Custom Streaming AG-UI Connection Model Request/Response Varies Server-Sent Events State Management Manual Manual Protocol-managed Tool Calling Invisible Custom format Standardized events Framework Varies Framework-locked Framework-agnostic Browser Support Universal Varies Universal Implementation Simple Complex Moderate Ecosystem N/A Isolated Growing You've now seen AG-UI's design principles, implementation details, and conceptual foundations. But the most important question remains: should you actually use it? Conclusion: Is AG-UI Right for Your Project? AG-UI represents a shift toward standardized, observable agent interactions. Before adopting it, understand where the protocol stands and whether it fits your needs. Protocol Maturity The protocol is stable enough for production use but still evolving: Ready now: Core specification stable, Microsoft Agent Framework integration available, FastAPI/Python implementation mature, basic streaming and threading work reliably. Choose AG-UI If You Building new agent projects - No legacy API to maintain, want future compatibility with emerging ecosystem Need streaming observability - Multi-step workflows where users benefit from seeing each stage of execution Want framework flexibility - Same client code works with any AG-UI-compliant backend Comfortable with evolving standards - Can adapt to protocol changes as it matures Stick with Alternatives If You Have working solutions - Custom streaming working well, migration cost not justified Need guaranteed stability - Mission-critical systems where breaking changes are unacceptable Build simple agents - Single-step request/response without tool calling or streaming needs Risk-averse environment - Large existing implementations where proven approaches are required Beyond individual project decisions, it's worth considering AG-UI's role in the broader ecosystem. The Bigger Picture While this blog post focused on Microsoft Agent Framework, AG-UI's true power lies in its broader mission: creating a common language for agent-UI communication across the entire ecosystem. As more frameworks adopt it, the real value emerges: write your UI once, work with any compliant agent framework. Think of it like GraphQL for APIs or OpenAPI for REST - a standardization layer that benefits the entire ecosystem. The protocol is young, but the problem it solves is real. Whether you adopt it now or wait for broader adoption, understanding AG-UI helps you make informed architectural decisions for your agent applications. Ready to dive deeper? Here are the official resources to continue your AG-UI journey. Resources AG-UI & Microsoft Agent Framework Getting Started with AG-UI (Microsoft Learn) - Official tutorial AG-UI Integration Overview - Architecture and concepts AG-UI Protocol Specification - Official protocol documentation Backend Tool Rendering - Adding function tools Security Considerations - Production security guidance Microsoft Agent Framework Documentation - Framework overview AG-UI Dojo Examples - Live demonstrations UI Components & Integration CopilotKit for Microsoft Agent Framework - React component library Community & Support Microsoft Q&A - Community support Agent Framework GitHub - Source code and issues Related Technologies Azure AI Foundry Documentation - Azure AI platform FastAPI Documentation - Web framework Server-Sent Events (SSE) Specification - Protocol standard This blog post introduces AG-UI with Microsoft Agent Framework, focusing on fundamental concepts and building your first interactive agent application.How to Build Safe Natural Language-Driven APIs
TL;DR Building production natural language APIs requires separating semantic parsing from execution. Use LLMs to translate user text into canonical structured requests (via schemas), then execute those requests deterministically. Key patterns: schema completion for clarification, confidence gates to prevent silent failures, code-based ontologies for normalization, and an orchestration layer. This keeps language as input, not as your API contract. Introduction APIs that accept natural language as input are quickly becoming the norm in the age of agentic AI apps and LLMs. From search and recommendations to workflows and automation, users increasingly expect to "just ask" and get results. But treating natural language as an API contract introduces serious risks in production systems: Nondeterministic behavior Prompt-driven business logic Difficult debugging and replay Silent failures that are hard to detect In this post, I'll describe a production-grade architecture for building safe, natural language-driven APIs: one that embraces LLMs for intent discovery and entity extraction while preserving the determinism, observability, and reliability that backend systems require. This approach is based on building real systems using Azure OpenAI and LangGraph, and on lessons learned the hard way. The Core Problem with Natural Language APIs Natural language is an excellent interface for humans. It is a poor interface for systems. When APIs accept raw text directly and execute logic based on it, several problems emerge: The API contract becomes implicit and unversioned Small prompt changes cause behavioral changes Business logic quietly migrates into prompts In short: language becomes the contract, and that's fragile. The solution is not to avoid natural language, but to contain it. A Key Principle: Natural Language Is Input, Not a Contract So how do we contain it? The answer lies in treating natural language fundamentally differently than we treat traditional API inputs. The most important design decision we made was this: Natural language should be translated into structure, not executed directly. That single principle drives the entire architecture. Instead of building "chatty APIs," we split responsibilities clearly: Natural language is used for intent discovery and entity extraction Structured data is used for execution Two Explicit API Layers This principle translates into a concrete architecture with two distinct API layers, each with a single, clear responsibility. 1. Semantic Parse API (Natural Language → Structure) This API: Accepts user text Extracts intent and entities using LLMs Completes a predefined schema Asks clarifying questions when required Returns a canonical, structured request Does not execute business logic Think of this as a compiler, not an engine. 2. Structured Execution API (Structure → Action) This API: Accepts only structured input Calls downstream systems to process the request and get results Is deterministic and versioned Contains no natural language handling Is fully testable and replayable This is where execution happens. Why This Separation Matters Separating these layers gives you: A stable, versionable API contract Freedom to improve NLP without breaking clients Clear ownership boundaries Deterministic execution paths Most importantly, it prevents LLM behavior from leaking into core business logic. Canonical Schemas Are the Backbone Now that we've established the two-layer architecture, let's dive into what makes it work: canonical schemas. Each supported intent is defined by a canonical schema that lives in code. Example (simplified): This schema is used when a user is looking for similar product recommendations. The entities capture which product to use as reference and how to bias the recommendations toward price or quality. { "intent": "recommend_similar", "entities": { "reference_product_id": "string", "price_bias": "number (-1 to 1)", "quality_bias": "number (-1 to 1)" } } Schemas define: Required vs optional fields Allowed ranges and types Validation rules They are the contract, not the prompt. When a user says "show me products like the blue backpack but cheaper", the LLM extracts: Intent: recommend_similar reference_product_id: "blue_backpack_123" price_bias: -0.8 (strongly prefer cheaper) quality_bias: 0.0 (neutral) The schema ensures that even if the user phrased it as "find alternatives to item 123 with better pricing" or "cheaper versions of that blue bag", the output is always the same structure. The natural language variation is absorbed at the semantic layer. The execution layer receives a consistent, validated request every time. This decoupling is what makes the system maintainable. Schema Completion, Not Free-Form Chat But what happens when the user's input doesn't contain all the information needed to complete the schema? This is where structured clarification comes in. A common misconception is that clarification means "chatting until it feels right." In production systems, clarification is schema completion. If required fields are missing or ambiguous, the semantic API responds with: What information is missing A targeted clarification question The current schema state Example response: { "status": "needs_clarification", "missing_fields": ["reference_product_id"], "question": "Which product should I compare against?", "state": { "intent": "recommend_similar", "entities": { "reference_product_id": null, "price_bias": -0.3, "quality_bias": 0.4 } } } The state object is the memory. The API itself remains stateless. A Complete Conversation Flow To illustrate how schema completion works in practice, here's a full conversation flow where the user's initial request is missing required information: Initial Request: User: "Show me cheaper alternatives with good quality" API Response (needs clarification): { "status": "needs_clarification", "missing_fields": ["reference_product_id"], "question": "Which product should I compare against?", "state": { "intent": "recommend_similar", "entities": { "reference_product_id": null, "price_bias": -0.3, "quality_bias": 0.4 } } } Follow-up Request: User: "The blue backpack" Client sends: { "user_input": "The blue backpack", "state": { "intent": "recommend_similar", "entities": { "reference_product_id": null, "price_bias": -0.3, "quality_bias": 0.4 } } } API Response (complete): { "status": "complete", "canonical_request": { "intent": "recommend_similar", "entities": { "reference_product_id": "blue_backpack_123", "price_bias": -0.3, "quality_bias": 0.4 } } } The client passes the state back with each clarification. The API remains stateless, while the client manages the conversation context. Once complete, the canonical_request can be sent directly to the execution API. Why LangGraph Fits This Problem Perfectly With schemas and clarification flows defined, we need a way to orchestrate the semantic parsing workflow reliably. This is where LangGraph becomes valuable. LangGraph allows semantic parsing to be modeled as a structured, deterministic workflow with explicit decision points: Classify intent: Determine what the user wants to do from a predefined set of supported actions Extract candidate entities: Pull out relevant parameters from the natural language input using the LLM Merge into schema state: Map the extracted values into the canonical schema structure Validate required fields: Check if all mandatory fields are present and values are within acceptable ranges Either complete or request clarification: Return the canonical request if complete, or ask a targeted question if information is missing Each node has a single responsibility. Validation and routing are done in code, not by the LLM. LangGraph provides: Explicit state transitions Deterministic routing Observable execution Safe retries Used this way, it becomes a powerful orchestration tool, not a conversational agent. Confidence Gates Prevent Silent Failures Structured workflows handle the process, but there's another critical safety mechanism we need: knowing when the LLM isn't confident about its extraction. Even when outputs are structurally valid, they may not be reliable. We require the semantic layer to emit a confidence score. If confidence falls below a threshold, execution is blocked and clarification is requested. This simple rule eliminates an entire class of silent misinterpretations that are otherwise very hard to detect. Example: When a user says "Show me items similar to the bag", the LLM might extract: { "intent": "recommend_similar", "confidence": 0.55, "entities": { "reference_product_id": "generic_bag_001", "confidence_scores": { "reference_product_id": 0.4 } } } The overall confidence is low (0.55), and the entity confidence for reference_product_id is very low (0.4) because "the bag" is ambiguous. There might be hundreds of bags in the catalog. Instead of proceeding with a potentially wrong guess, the API responds: { "status": "needs_clarification", "reason": "low_confidence", "question": "I found multiple bags. Did you mean the blue backpack, the leather tote, or the travel duffel?", "confidence": 0.55 } This prevents the system from silently executing the wrong recommendation and provides a better user experience. Lightweight Ontologies (Keep Them in Code) Beyond confidence scoring, we need a way to normalize the variety of terms users might use into consistent canonical values. We also introduced lightweight, code-level ontologies: Allowed intents Required entities per intent Synonym-to-canonical mappings Cross-field validation rules These live in code and configuration, not in prompts. LLMs propose values. Code enforces meaning. Example: Consider these user inputs that all mean the same thing: "Show me cheaper options" "Find budget-friendly alternatives" "I want something more affordable" "Give me lower-priced items" The LLM might extract different values: "cheaper", "budget-friendly", "affordable", "lower-priced". The ontology maps all of these to a canonical value: PRICE_BIAS_SYNONYMS = { "cheaper": -0.7, "budget-friendly": -0.7, "affordable": -0.7, "lower-priced": -0.7, "expensive": 0.7, "premium": 0.7, "high-end": 0.7 } When the LLM extracts "budget-friendly", the code normalizes it to -0.7 for the price_bias field. Similarly, cross-field validation catches logical inconsistencies: if entities["price_bias"] < -0.5 and entities["quality_bias"] > 0.5: return clarification("You want cheaper items with higher quality. This might be difficult. Should I prioritize price or quality?") The LLM proposes. The ontology normalizes. The validation enforces business rules. What About Latency? A common concern with multi-step semantic parsing is performance. In practice, we observed: Intent classification: ~40 ms Entity extraction: ~200 ms Validation and routing: ~1 ms Total overhead: ~250–300 ms. For chat-driven user experiences, this is well within acceptable bounds and far cheaper than incorrect or inconsistent execution. Key Takeaways Let's bring it all together. If you're building APIs that accept natural language in production: Do not make language your API contract Translate language into canonical structure Own schema completion server-side Use LLMs for discovery and extraction, not execution Treat safety and determinism as first-class requirements Natural language is an input format. Structure is the contract. Closing Thoughts LLMs make it easy to build impressive demos. Building safe, reliable systems with them requires discipline. By separating semantic interpretation from execution, and by using tools like Azure OpenAI and LangGraph thoughtfully, you can build natural language-driven APIs that scale, evolve, and behave predictably in production. Hopefully, this architecture saves you a few painful iterations.Using on-behalf-of flow for Entra-based MCP servers
In December, we presented a series about MCP, culminating in a session about adding authentication to MCP servers. I demoed a Python MCP server that uses Microsoft Entra for authentication, requiring users to first login to the Microsoft tenant before they could use a tool. Many developers asked how they could take the Entra integration further, like to check the user's group membership or query their OneDrive. That requires using an "on-behalf-of" flow, where the MCP server uses the user's Entra identity to call another API, like the Microsoft Graph API. In this blog post, I will explain how to use Entra with OBO flow in a Python FastMCP server. How MCP servers can use Entra authentication The MCP authorization specification is based on OAuth2, with some additional features tacked on top. Every MCP client is actually an OAuth2 client, and each MCP server is an OAuth2 resource server: MCP auth adds these features to help clients determine how to authorize a server: Protected resource metadata (PRM): Implemented on the MCP server, provides details about the authorization server and method Authorization server metadata: Implemented on the authorization server, gives URLs for OAuth2 endpoints Additionally, to allow MCP servers to work with arbitrary MCP clients, MCP auth supports either of these client registration methods: Dynamic Client Registration (DCR): Implemented on the authorization server, it can register new MCP clients as OAuth2 clients, even if it hasn't seen them before. Client ID Metadata Documents (CIMD): An alternative to DCR, this requires both the MCP client to make a CIMD document available on a server, and requires the authorization server to fetch the CIMD document for details about the client. Microsoft Entra does support authorization server metadata, but it does not support either DCR or CIMD. That's actually fine if you are building an MCP server that's only going to be used with pre-authorized clients, like if the server will only be used with VS Code or with a specific internal MCP client. But, if you are building an MCP server that can be used with arbitrary MCP clients, then either DCR or CIMD is required. So what do we do? Fortunately, the FastMCP SDK implements DCR on top of Entra using an OAuth proxy pattern. FastMCP acts as the authorization server, intercepting requests and forwarding to Entra when needed, and storing OAuth client information in a designated database (like in-memory or Cosmos DB). ⚠️ Warning: This proxy approach is intended only for development and testing scenarios. For production deployments, Microsoft recommends using pre‑registered client applications where client identifiers and permissions are explicitly created, reviewed, and approved on a per-app basis. Let's walk through the steps to set that up. Registering the server with Entra Before the server can use Entra to authorize users, we need to register the server with Entra via an app registration. We can do registration using the Azure Portal, Azure CLI, Microsoft Graph SDK, or even Bicep. In this demo, I use the Python MS Graph SDK as it allows me to specify everything programmatically. First, I create the Entra app registration, specifying the sign-in audience (single-tenant), redirect URIs (including local MCP server, deployed MCP server, and VS Code redirect URIs), and the scopes for the exposed API. request_app = Application( display_name="FastMCP Server App", sign_in_audience="AzureADMyOrg", # Single tenant web=WebApplication( redirect_uris=[ "http://localhost:8000/auth/callback", "https://vscode.dev/redirect", "http://127.0.0.1:33418", "https://deployedurl.com/auth/callback" ], ), api=ApiApplication( oauth2_permission_scopes=[ PermissionScope( id=uuid.UUID("{" + str(uuid.uuid4()) + "}"), admin_consent_display_name="Access FastMCP Server", admin_consent_description="Allows access to the FastMCP server as the signed-in user.", user_consent_display_name="Access FastMCP Server", user_consent_description="Allow access to the FastMCP server on your behalf", is_enabled=True, value="mcp-access", type="User", )], requested_access_token_version=2, # Required by FastMCP ) ) app = await graph_client.applications.post(request_app) await graph_client.applications.by_application_id(app.id).patch( Application(identifier_uris=[f"api://{app.app_id}"])) Thanks to that configuration, when an MCP client like VS Code requests an OAuth2 token, it will request a token with the scope "api://{app.app_id}/mcp-access", and the FastMCP server will validate that incoming tokens contain that scope. Next, I create a Service Principal for that Entra app registration, which represents the Entra app in my tenant request_principal = ServicePrincipal(app_id=app.app_id, display_name=app.display_name) await graph_client.service_principals.post(request_principal) I need a way for the server to prove that it can use that Entra app registration, so I register a secret: password_credential = await graph_client.applications.by_application_id(app.id).add_password.post( AddPasswordPostRequestBody( password_credential=PasswordCredential(display_name="FastMCPSecret"))) Ideally, I would like to move away from secrets, as Entra now has support for using federated identity credentials for Entra app registrations instead, but that form of credential isn't supported yet in the FastMCP SDK. If you choose to use a secret, make sure that you store the secret securely. Granting admin consent This next step is only necessary when our MCP server wants to use an OBO flow to exchange access tokens for other resource server tokens (Graph API tokens, in this case). For the OBO flow to work, the Entra app registration needs permission to call the Graph API on behalf of users. If we controlled the client, we could force it to request the required scopes as part of the initial login dialog. However, since we are configuring this server to work with arbitrary MCP clients, we don't have that option. Instead, we grant admin consent to the Entra app for the necessary scopes, such that no Graph API consent dialog is needed. This code grants admin consent to the associated service principal for the Graph API resource and scopes: server_principal = await graph_client.service_principals_with_app_id(app.app_id).get() grant = GrantDefinition( principal_id=server_principal.id, resource_app_id="00000003-0000-0000-c000-000000000000", # Graph API scopes=["User.Read", "email", "offline_access", "openid", "profile"], target_label="server application") resource_principal = await graph_client.service_principals_with_app_id(grant.resource_app_id).get() desired_scope = grant.scope_string() await graph_client.oauth2_permission_grants.post( OAuth2PermissionGrant( client_id=grant.principal_id, consent_type="AllPrincipals", resource_id=resource_principal.id, scope=desired_scope)) If our MCP server needed to use an OBO flow with another resource server, we could request additional grants for those resources and scopes. Our Entra app registration is now ready for the MCP server, so let's move on to see the server code. Using FastMCP servers with Entra In our MCP server code, we configure FastMCP's built in AzureProvider based off the details from the Entra app registration process: auth = AzureProvider( client_id=os.environ["ENTRA_PROXY_AZURE_CLIENT_ID"], client_secret=os.environ["ENTRA_PROXY_AZURE_CLIENT_SECRET"], tenant_id=os.environ["AZURE_TENANT_ID"], base_url=entra_base_url, # MCP server URL required_scopes=["mcp-access"], client_storage=oauth_client_store, # in-memory or Cosmos DB ) To make it easy for our MCP tools to access an identifier for the currently logged in user, we define a middleware that inspects the claims of the current token using FastMCP's get_access_token() and sets the "oid" (Entra object identifier) in the state: class UserAuthMiddleware(Middleware): def _get_user_id(self): token = get_access_token() if not (token and hasattr(token, "claims")): return None return token.claims.get("oid") async def on_call_tool(self, context: MiddlewareContext, call_next): user_id = self._get_user_id() if context.fastmcp_context is not None: context.fastmcp_context.set_state("user_id", user_id) return await call_next(context) async def on_read_resource(self, context: MiddlewareContext, call_next): user_id = self._get_user_id() if context.fastmcp_context is not None: context.fastmcp_context.set_state("user_id", user_id) return await call_next(context) When we initialize the FastMCP server, we set the auth provider and include that middleware: mcp = FastMCP("Expenses Tracker", auth=auth, middleware=[UserAuthMiddleware()]) Now, every request made to the MCP server will require authentication. The server will return a 401 if a valid token isn't provided, and that 401 will prompt the MCP client to kick off the MCP authorization flow. Inside each tool, we can grab the user id from the state, and use that to customize the response for the user, like to store or query items in a database. .tool async def add_user_expense( date: Annotated[date, "Date of the expense in YYYY-MM-DD format"], amount: Annotated[float, "Positive numeric amount of the expense"], description: Annotated[str, "Human-readable description of the expense"], ctx: Context, ): """Add a new expense to Cosmos DB.""" user_id = ctx.get_state("user_id") if not user_id: return "Error: Authentication required (no user_id present)" expense_item = { "id": str(uuid.uuid4()), "user_id": user_id, "date": date.isoformat(), "amount": amount, "description": description } await cosmos_container.create_item(body=expense_item) Using OBO flow in FastMCP server Now we can move on to using an OBO flow inside an MCP tool, to access the Graph API on behalf of the user. To make it easy to exchange Entra tokens for Graph tokens, we use the Python MSAL SDK, configuring a ConfidentialClientApplication based on our Entra app registration details: confidential_client = ConfidentialClientApplication( client_id=os.environ["ENTRA_PROXY_AZURE_CLIENT_ID"], client_credential=os.environ["ENTRA_PROXY_AZURE_CLIENT_SECRET"], authority=f"https://login.microsoftonline.com/{os.environ['AZURE_TENANT_ID']}", token_cache=TokenCache(), ) Inside the tool that requires OBO, we ask MSAL to exchange the MCP access token for a Graph API access token: access_token = get_access_token() graph_resource_access_token = confidential_client.acquire_token_on_behalf_of( user_assertion=access_token.token, scopes=["https://graph.microsoft.com/.default"] ) graph_token = graph_resource_access_token["access_token"] Once we successfully acquire the token, we can use that token with the Graph API, for any operations permitted by the scopes in the admin consent granted earlier. For this example, we call the Graph API to check whether the logged in user is a member of a particular Entra group, and restrict tool usage if not: async with httpx.AsyncClient() as client: url = ("https://graph.microsoft.com/v1.0/me/transitiveMemberOf/microsoft.graph.group" f"?$filter=id eq '{group_id}'&$count=true") response = await client.get( url, headers={ "Authorization": f"Bearer {graph_token}", "ConsistencyLevel": "eventual", }) data = response.json() membership_count = data.get("@odata.count", 0) You could imagine many other ways to use an OBO flow however, like to query for more details from the Graph API, upload documents to OneDrive/SharePoint/Notes, send emails, and more! All together now For the full code, check out the open source python-mcp-demos repository, and follow the deployment steps for Entra. The most relevant code files are: auth_init.py: Creates the Entra app registration, service principal, client secret, and grants admin consent for OBO flow. auth_update.py: Updates the app registration's redirect URIs after deployment, adding the deployed server URL. auth_entra_mcp.py: The MCP server itself, configured with FastMCP's AzureProvider and tools that use OBO for group membership checks. I want to reiterate once more that the OAuth proxy approach is intended only for development and testing scenarios. For production deployments, Microsoft recommends using pre‑registered client applications where client identifiers and permissions are explicitly created, reviewed, and approved on a per-app basis. I hope that in the future, Entra will formally support MCP authorization via the CIMD protocol, so that we can build MCP servers with Entra auth that work with MCP clients in a fully secure and production-ready way. As always, please let us know if you have further questions or ideas for other Entra integrations. Acknowledgements: Thank you to Matt Gotteiner for his guidance in implementing the OBO flow and review of the blog post.Join our free livestream series on building agents in Python
Join us for a new 6‑part livestream series where we explore the foundational concepts behind building AI agents in Python using the Microsoft Agent Framework. This series is for anyone who wants to understand how agents work—how they call tools, use memory and context, and construct workflows on top of them. Over two weeks, we’ll dive into the practical building blocks that shape real agent behavior. You’ll learn how to: 🔧 Register and structure tools 🔗 Connect local MCP servers 📚 Add context with database calls 🧠 Add memory for personalization 📈 Monitor agent behavior with OpenTelemetry ✅ Evaluate the quality of agent output Throughout the series, we’ll use Python for all live examples and share full code so you can run everything yourself. You can also follow along live using GitHub Models and GitHub Codespaces. 👉 Register for the full series. Spanish speaker? ¡Tendremos una serie para hispanohablantes! Regístrese aquí In addition to the live streams, you can also join Join the Microsoft Foundry Discord to ask follow-up questions after each stream. If you are brand new to generative AI with Python, start with our 9-part Python + AI series, which covers topics such as LLMs, embedding models, RAG, tool calling, MCP, and will prepare you perfectly for the agents series. To learn more about each live stream or register for individual sessions, scroll down: Python + Agents: Building your first agent in Python 24 February, 2026 | 6:30 PM - 7:30 PM (UTC) Coordinated Universal Time Register for the stream on Reactor In the first session of our Python + Agents series, we’ll kick things off with the fundamentals: what AI agents are, how they work, and how to build your first one using the Microsoft Agent Framework. We’ll start with the core anatomy of an agent, then walk through how tool calling works in practice—beginning with a single tool, expanding to multiple tools, and finally connecting to tools exposed through local MCP servers. We’ll conclude with the supervisor agent pattern, where a single supervisor agent coordinates subtasks across multiple subagents, by treating each agent as a tool. Along the way, we'll share tips for debugging and inspecting agents, like using the DevUI interface from Microsoft Agent Framework for interacting with agent prototypes. Python + Agents: Adding context and memory to agents 25 February, 2026 | 6:30 PM - 7:30 PM (UTC) Coordinated Universal Time Register for the stream on Reactor In the second session of our Python + Agents series, we’ll extend agents built with the Microsoft Agent Framework by adding two essential capabilities: context and memory. We’ll begin with context, commonly known as Retrieval‑Augmented Generation (RAG), and show how agents can ground their responses using knowledge retrieved from local data sources such as SQLite or PostgreSQL. This enables agents to provide accurate, domain‑specific answers based on real information rather than model hallucination. Next, we’ll explore memory—both short‑term, thread‑level context and long‑term, persistent memory. You’ll see how agents can store and recall information using solutions like Redis or open‑source libraries such as Mem0, enabling them to remember previous interactions, user preferences, and evolving tasks across sessions. By the end, you’ll understand how to build agents that are not only capable but context‑aware and memory‑efficient, resulting in richer, more personalized user experiences. Python + Agents: Monitoring and evaluating agents 26 February, 2026 | 6:30 PM - 7:30 PM (UTC) Coordinated Universal Time Register for the stream on Reactor In the third session of our Python + Agents series, we’ll focus on two essential components of building reliable agents: observability and evaluation. We’ll begin with observability, using OpenTelemetry to capture traces, metrics, and logs from agent actions. You'll learn how to instrument your agents and use a local Aspire dashboard to identify slowdowns and failures. From there, we’ll explore how to evaluate agent behavior using the Azure AI Evaluation SDK. You’ll see how to define evaluation criteria, run automated assessments over a set of tasks, and analyze the results to measure accuracy, helpfulness, and task success. By the end of the session, you’ll have practical tools and workflows for monitoring, measuring, and improving your agents—so they’re not just functional, but dependable and verifiably effective. Python + Agents: Building your first AI-driven workflows 3 March, 2026 | 6:30 PM - 7:30 PM (UTC) Coordinated Universal Time Register for the stream on Reactor In Session 4 of our Python + Agents series, we’ll explore the foundations of building AI‑driven workflows using the Microsoft Agent Framework: defining workflow steps, connecting them, passing data between them, and introducing simple ways to guide the path a workflow takes. We’ll begin with a conceptual overview of workflows and walk through their core components: executors, edges, and events. You’ll learn how workflows can be composed of simple Python functions or powered by full AI agents when a step requires model‑driven behavior. From there, we’ll dig into conditional branching, showing how workflows can follow different paths depending on model outputs, intermediate results, or lightweight decision functions. We’ll introduce structured outputs as a way to make branching more reliable and easier to maintain—avoiding vague string checks and ensuring that workflow decisions are based on clear, typed data. We'll discover how the DevUI interface makes it easier to develop workflows by visualizing the workflow graph and surfacing the streaming events during a workflow's execution. Finally, we'll dive into an E2E demo application that uses workflows inside a user-facing application with a frontend and backend. Python + Agents: Orchestrating advanced multi-agent workflows 4 March, 2026 | 6:30 PM - 7:30 PM (UTC) Coordinated Universal Time Register for the stream on Reactor In Session 5 of our Python + Agents series, we’ll go beyond workflow fundamentals and explore how to orchestrate advanced, multi‑agent workflows using the Microsoft Agent Framework. This session focuses on patterns that coordinate multiple steps or multiple agents at once, enabling more powerful and flexible AI‑driven systems. We’ll begin by comparing sequential vs. concurrent execution, then dive into techniques for running workflow steps in parallel. You’ll learn how fan‑out and fan‑in edges enable multiple branches to run at the same time, how to aggregate their results, and how concurrency allows workflows to scale across tasks efficiently. From there, we’ll introduce two multi‑agent orchestration approaches that are built into the framework. We’ll start with handoff, where control moves entirely from one agent to another based on workflow logic, which is useful for routing tasks to the right agent as the workflow progresses. We’ll then look at Magentic, a planning‑oriented supervisor that generates a high‑level plan for completing a task and delegates portions of that plan to other agents. Finally, we'll wrap up with a demo of an E2E application that showcases a concurrent multi-agent workflow in action. Python + Agents: Adding a human in the loop to agentic workflows 5 March, 2026 | 6:30 PM - 7:30 PM (UTC) Coordinated Universal Time Register for the stream on Reactor In the final session of our Python + Agents series, we’ll explore how to incorporate human‑in‑the‑loop (HITL) interactions into agentic workflows using the Microsoft Agent Framework. This session focuses on adding points where a workflow can pause, request input or approval from a user, and then resume once the human has responded. HITL is especially important because LLMs can produce uncertain or inconsistent outputs, and human checkpoints provide an added layer of accuracy and oversight. We’ll begin with the framework’s requests‑and‑responses model, which provides a structured way for workflows to ask questions, collect human input, and continue execution with that data. We'll move onto tool approval, one of the most frequent reasons an agent requests input from a human, and see how workflows can surface pending tool calls for approval or rejection. Next, we’ll cover checkpoints and resuming, which allow workflows to pause and be restarted later. This is especially important for HITL scenarios where the human may not be available immediately. We’ll walk through examples that demonstrate how checkpoints store progress, how resuming picks up the workflow state, and how this mechanism supports longer‑running or multi‑step review cycles. This session brings together everything from the series—agents, workflows, branching, orchestration—and shows how to integrate humans thoughtfully into AI‑driven processes, especially when reliability and judgment matter most.Data Security: Azure key Vault in Data bricks
Why this article? To remove the vulnerability of exposing the data base connection string in Databricks notebook directly, by using Azure key vault. Database connection strings are extremely confidential/vulnerable data, that we should not be exposed in the DataBricks notebook explicitly. Azure key vault is a secure option to read the secrets and establish connection. What do we need? Tenant Id of the app from the app registration with access to the azure key vault secrets Client Id of the of the app from the app registration with access to the azure key vault secrets Client secret of the app from the app registration with access to the azure key vault Where to find this information? Under the App registration, you can find the (application) Client Id, Directory (tenant) Id. Client secret value is found in the app registration of the service, under Manage -> Certificate & secrets. You can use an existing secret or create a new one and use it to access the key Vault secrets. Make sure the application is added with get access to read the secrets. Verify the key vault you are checking and using in Databricks is the same one with read access. You can verify this by going to the Azure key vault -> Access Policies and search for the application name. It should show up on search as below, this will confirm that the access of the application. What do we need to setup in Databricks notebook? Open your cluster and install azure.keyvault and azure-identity (installing version should be compatible with you cluster configuration, refer: https://docs.databricks.com/aws/en/libraries/package-repositories) In a new notebook, let’s start by importing the necessary modules. Your notebook would start with the modules, followed by tentatId, clientId, client secret, azure key vault URL , secretName of the connection string in the azure key vault and secretVersion. Lastly, we need to fetch the secret using the below code Vola, we have the DB connection string to perform the CRUD operations. Conclusion: By securely retrieving your database connection string from Azure Key Vault, you eliminate credential exposure and strengthen the overall security posture of your Databricks workflows. This simple shift ensures your notebooks remain clean, compliant, and production‑ready.Host ChatGPT apps on Azure Functions
This blog post is for developers learning and building ChatGPT apps. It provides an overview of how these apps work, why build them, and how to host one on Azure Functions. Chat with ChatGPT apps OpenAI recently launched ChatGPT apps. These are apps you can chat with right inside ChatGPT, extending what ChatGPT can do beyond simple chats to actions. These apps can be invoked by starting a message with the app name, or they can be suggested by ChatGPT when relevant to the conversation. The following shows an example of invoking the Booking.com app to find hotels that meet certain criteria: OpenAI calls these “a new generation of apps” that “blend familiar interactive elements…with new ways of interacting through conversation.” For users, ChatGPT apps fit directly into an interface they’re already familiar with and can use with little to no learning. For developers, building these apps is great way to get them in the hands of ChatGPT’s 800 million users without having to build custom frontends or worry about distribution and discovery. The following summarizes key benefits of ChatGPT apps: Native Integration: Once connected, users can invoke apps with a simple @ mention. Contextual Actions: Your app doesn't just "chat"—it does. It can fetch real-time data or execute actions. Massive Distribution and Easy Discovery: ChatGPT has added an app directory and just announced that they’re accepting submissions. Apps in the directory are exposed ChatGPT’s massive user base. ChatGPT apps are remote MCP servers ChatGPT apps are simply remote MCP servers that expose tools, but with two notable distinctions: Their tools use metadata to specify UI elements that should be rendered when it returns a result The UI elements are exposed as MCP resources. ChatGPT invokes tools the same way agents invoke tools on any MCP server. The difference is the added ability to render the tool results in a custom UI that’s embedded in the chat as an iFrame. A UI can include buttons, text boxes, maps, and other components that users can interact with. Instead of calling a RESTful API, the UI can trigger additional tool calls in the MCP server as the user interacts with it. Learn more about building the custom UI . For example, when the Zillow app returns results to the user’s question, the results are home listings and a map that users can interact with: Since ChatGPT apps are just MCP servers, any existing server you may have can be turned into a ChatGPT app. To do that, you must ensure the server uses the streamable HTTP transport if it doesn’t already and then find a place to host it. Hosting remote MCP servers While there are many hosting platforms available, Azure Functions is uniquely positioned to host remote MCP servers as the platform provides several key benefits: Scalable Infrastructure: ChatGPT apps can go viral. Azure Function’s Flex Consumption plan can handle bursty traffic during high traffic times and scale back to zero when needed Built-in auth: Keep your server secured with Azure Function’s built-in server authentication and authorization feature Serverless billing: Pay for only when the app is run instead of idle time Learn more about remote MCP servers hosted on Azure Functions. Create ChatGPT app Let’s quickly create a sample ChatGPT app that returns the weather of a place. Prerequisites Ensure you have the following prerequisites before proceeding: Azure subscription for creating Azure Functions and related resources Azure Developer CLI for deploying MCP server via infrastructure as code ChatGPT Plus subscription for testing ChatGPT app in developer mode Deploy MCP server to Azure Functions Clone this sample MCP server: `git clone https://github.com/Azure-Samples/chatgpt-app-azure-function-mcp`. Open terminal, run `azd auth login` and complete the login flow in the browser. Navigate to sample root directory, run `azd up` to deploy the server and related resources. You’ll be prompted with: Enter a unique environment name: Enter a unique name. This is the name of the resource group where all deployed resources live. Select an Azure Subscription: Pick your subscription Enter a value for the ‘location’ infrastructure: East US Once deployment completes, copy the app url for the next step. It should look like: https://<your-app>.azurewebsites.net Sample code walkthrough The sample server is built using the Python FastMCP package. You can find more information and how to test server locally in this repo. We'll walkthough the code briefly here. In main.py, you find the `get_weather_widget` resource and `get_current_weather` tool (code abbreviated here): .resource("ui://widget/current-weather.html", mime_type="text/html+skybridge") def get_weather_widget() -> str: """Interactive HTML widget to display current weather data in ChatGPT.""" # some code... @mcp.tool( annotations={ "title": "Get Current Weather", "readOnlyHint": True, "openWorldHint": True, }, meta={ "openai/outputTemplate": "ui://widget/current-weather.html", "openai/toolInvocation/invoking": "Fetching weather data", "openai/toolInvocation/invoked": "Weather data retrieved" }, ) def get_current_weather(latitude: float, longitude: float) -> ToolResult: """Get current weather for a given latitude and longitude using Open-Meteo API.""" # some code... return ToolResult( content=content_text, structured_content=data ) When you ask ChatGPT a question, it calls the MCP tool which returns a `ToolResult` containing both human-readable content (for ChatGPT to understand) and machine-readable data (`structured_content`, raw data for the widget). Because the `get_current_weather` tool specifies an `outputTemplate` in the metadata, ChatGPT fetches the corresponding widget HTML from the `get_weather_widget` resource. To return results, it creates an iframe and injects the weather results (`structured_content`) into the widget's JavaScript environment (via `window.openai.toolOutput`). The widget's JavaScript then renders the weather data into a beautiful UI. Test ChatGPT app in developer mode Turn on Developer mode in ChatGPT: Go to Settings → Connectors → Advanced → Developer mode In the chat, click + → More → Add sources The Add + button should show next to Sources. Click Add + → Connect more In the Enable apps window, look for Advanced settings. Click Create app. A form should open. Fill out the form to create the new app Name: WeatherApp MCP Server URL: Enter the MCP server endpoint, which is the app URL you previously saved with /mcp appended. Example: https://<you-app>.azurewebsites.net/mcp Authentication: Choose No Auth Check the box for “I understand and want to continue” and click Create. Once connected, you should find the server listed under Enabled apps. Test by asking ChatGPT “@WeatherApp what’s the temperature in NYC today?” Submit to ChatGPT App Directory OpenAI has opened app submission recently. Submitting the app to the App Directory makes it accessible to all users on ChatGPT. You may want to read through the submission guidelines to ensure your app meets the requirements before submitting. What’s next In this blog post, we gave an overview of ChatGPT apps and showed how to host one in Azure Functions. We’ll dedicate the next blog post to elaborate on configuring authentication and authorization for apps hosted on Azure Functions. For users familiar with the Azure Functions MCP extension, we’re working on support for MCP Resources in the extension. You’ll be able to build ChatGPT apps using the extension once that support is out. For now, you need to use the official MCP SDKs. Closing thoughts ChatGPT apps extend the ability of ChatGPT beyond chat by letting users take actions like searching for an apartment, ordering groceries, and turning an outline into slide deck with just a mention of the app name in the chat. The directory OpenAI created where developers can submit their apps reminds one of the App Store in the iPhone. It seems to be a no-brainer now that such a marketplace should be provided. Would this also be the case for ChatGPT? Do you think the introduction of these apps is a gamechanger? And are they useful for your scenarios? Share with us your thoughts!Learn how to build MCP servers with Python and Azure
We just concluded Python + MCP, a three-part livestream series where we: Built MCP servers in Python using FastMCP Deployed them into production on Azure (Container Apps and Functions) Added authentication, including Microsoft Entra as the OAuth provider All of the materials from our series are available for you to keep learning from, and linked below: Video recordings of each stream Powerpoint slides Open-source code samples complete with Azure infrastructure and 1-command deployment If you're an instructor, feel free to use the slides and code examples in your own classes. Spanish speaker? We've got you covered- check out the Spanish version of the series. 🙋🏽♂️Have follow up questions? Join our weekly office hours on Foundry Discord: Tuesdays @ 11AM PT → Python + AI Thursdays @ 8:30 AM PT → All things MCP Building MCP servers with FastMCP 📺 Watch YouTube recording In the intro session of our Python + MCP series, we dive into the hottest technology of 2025: MCP (Model Context Protocol). This open protocol makes it easy to extend AI agents and chatbots with custom functionality, making them more powerful and flexible. We demonstrate how to use the Python FastMCP SDK to build an MCP server running locally. Then we consume that server from chatbots like GitHub Copilot in VS Code, using it's tools, resources, and prompts. Finally, we discover how easy it is to connect AI agent frameworks like Langchain and Microsoft agent-framework to the MCP server. Slides for this session Code repository with examples: python-mcp-demos Deploying MCP servers to the cloud 📺 Watch YouTube recording In our second session of the Python + MCP series, we deploy MCP servers to the cloud! We walk through the process of containerizing a FastMCP server with Docker and deploying to Azure Container Apps. Then we instrument the MCP server with OpenTelemetry and observe the tool calls using Azure Application Insights and Logfire. Finally, we explore private networking options for MCP servers, using virtual networks that restrict external access to internal MCP tools and agents. Slides for this session Code repository with examples: python-mcp-demos Authentication for MCP servers 📺 Watch YouTube recording In our third session of the Python + MCP series, we explore the best ways to build authentication layers on top of your MCP servers. We start off simple, with an API key to gate access, and demonstrate a key-restricted FastMCP server deployed to Azure Functions. Then we move on to OAuth-based authentication for MCP servers that provide user-specific data. We dive deep into MCP authentication, which is built on top of OAuth2 but with additional requirements like PRM and DCR/CIMD, which can make it difficult to implement fully. We demonstrate the full MCP auth flow in the open-souce identity provider KeyCloak, and show how to use an OAuth proxy pattern to implement MCP auth on top of Microsoft Entra. Slides for this session Code repository with Container Apps examples: python-mcp-demos Code repository with Functions examples: python-mcp-demosFrom Cloud to Chip: Building Smarter AI at the Edge with Windows AI PCs
As AI engineers, we’ve spent years optimizing models for the cloud, scaling inference, wrangling latency, and chasing compute across clusters. But the frontier is shifting. With the rise of Windows AI PCs and powerful local accelerators, the edge is no longer a constraint it’s now a canvas. Whether you're deploying vision models to industrial cameras, optimizing speech interfaces for offline assistants, or building privacy-preserving apps for healthcare, Edge AI is where real-world intelligence meets real-time performance. Why Edge AI, Why Now? Edge AI isn’t just about running models locally, it’s about rethinking the entire lifecycle: - Latency: Decisions in milliseconds, not round-trips to the cloud. - Privacy: Sensitive data stays on-device, enabling HIPAA/GDPR compliance. - Resilience: Offline-first apps that don’t break when the network does. - Cost: Reduced cloud compute and bandwidth overhead. With Windows AI PCs powered by Intel and Qualcomm NPUs and tools like ONNX Runtime, DirectML, and Olive, developers can now optimize and deploy models with unprecedented efficiency. What You’ll Learn in Edge AI for Beginners The Edge AI for Beginners curriculum is a hands-on, open-source guide designed for engineers ready to move from theory to deployment. Multi-Language Support This content is available in over 48 languages, so you can read and study in your native language. What You'll Master This course takes you from fundamental concepts to production-ready implementations, covering: Small Language Models (SLMs) optimized for edge deployment Hardware-aware optimization across diverse platforms Real-time inference with privacy-preserving capabilities Production deployment strategies for enterprise applications Why EdgeAI Matters Edge AI represents a paradigm shift that addresses critical modern challenges: Privacy & Security: Process sensitive data locally without cloud exposure Real-time Performance: Eliminate network latency for time-critical applications Cost Efficiency: Reduce bandwidth and cloud computing expenses Resilient Operations: Maintain functionality during network outages Regulatory Compliance: Meet data sovereignty requirements Edge AI Edge AI refers to running AI algorithms and language models locally on hardware, close to where data is generated without relying on cloud resources for inference. It reduces latency, enhances privacy, and enables real-time decision-making. Core Principles: On-device inference: AI models run on edge devices (phones, routers, microcontrollers, industrial PCs) Offline capability: Functions without persistent internet connectivity Low latency: Immediate responses suited for real-time systems Data sovereignty: Keeps sensitive data local, improving security and compliance Small Language Models (SLMs) SLMs like Phi-4, Mistral-7B, Qwen and Gemma are optimized versions of larger LLMs, trained or distilled for: Reduced memory footprint: Efficient use of limited edge device memory Lower compute demand: Optimized for CPU and edge GPU performance Faster startup times: Quick initialization for responsive applications They unlock powerful NLP capabilities while meeting the constraints of: Embedded systems: IoT devices and industrial controllers Mobile devices: Smartphones and tablets with offline capabilities IoT Devices: Sensors and smart devices with limited resources Edge servers: Local processing units with limited GPU resources Personal Computers: Desktop and laptop deployment scenarios Course Modules & Navigation Course duration. 10 hours of content Module Topic Focus Area Key Content Level Duration 📖 00 Introduction to EdgeAI Foundation & Context EdgeAI Overview • Industry Applications • SLM Introduction • Learning Objectives Beginner 1-2 hrs 📚 01 EdgeAI Fundamentals Cloud vs Edge AI comparison EdgeAI Fundamentals • Real World Case Studies • Implementation Guide • Edge Deployment Beginner 3-4 hrs 🧠 02 SLM Model Foundations Model families & architecture Phi Family • Qwen Family • Gemma Family • BitNET • μModel • Phi-Silica Beginner 4-5 hrs 🚀 03 SLM Deployment Practice Local & cloud deployment Advanced Learning • Local Environment • Cloud Deployment Intermediate 4-5 hrs ⚙️ 04 Model Optimization Toolkit Cross-platform optimization Introduction • Llama.cpp • Microsoft Olive • OpenVINO • Apple MLX • Workflow Synthesis Intermediate 5-6 hrs 🔧 05 SLMOps Production Production operations SLMOps Introduction • Model Distillation • Fine-tuning • Production Deployment Advanced 5-6 hrs 🤖 06 AI Agents & Function Calling Agent frameworks & MCP Agent Introduction • Function Calling • Model Context Protocol Advanced 4-5 hrs 💻 07 Platform Implementation Cross-platform samples AI Toolkit • Foundry Local • Windows Development Advanced 3-4 hrs 🏭 08 Foundry Local Toolkit Production-ready samples Sample applications (see details below) Expert 8-10 hrs Each module includes Jupyter notebooks, code samples, and deployment walkthroughs, perfect for engineers who learn by doing. Developer Highlights - 🔧 Olive: Microsoft's optimization toolchain for quantization, pruning, and acceleration. - 🧩 ONNX Runtime: Cross-platform inference engine with support for CPU, GPU, and NPU. - 🎮 DirectML: GPU-accelerated ML API for Windows, ideal for gaming and real-time apps. - 🖥️ Windows AI PCs: Devices with built-in NPUs for low-power, high-performance inference. Local AI: Beyond the Edge Local AI isn’t just about inference, it’s about autonomy. Imagine agents that: - Learn from local context - Adapt to user behavior - Respect privacy by design With tools like Agent Framework, Azure AI Foundry and Windows Copilot Studio, and Foundry Local developers can orchestrate local agents that blend LLMs, sensors, and user preferences, all without cloud dependency. Try It Yourself Ready to get started? Clone the Edge AI for Beginners GitHub repo, run the notebooks, and deploy your first model to a Windows AI PC or IoT devices Whether you're building smart kiosks, offline assistants, or industrial monitors, this curriculum gives you the scaffolding to go from prototype to production.
Unlocking Application Modernisation with GitHub Copilot
AI-driven modernisation is unlocking new opportunities you may not have even considered yet. It's also allowing organisations to re-evaluate previously discarded modernisation attempts that were considered too hard, complex or simply didn't have the skills or time to do. During Microsoft Build 2025, we were introduced to the concept of Agentic AI modernisation and this post from Ikenna Okeke does a great job of summarising the topic - Reimagining App Modernisation for the Era of AI | Microsoft Community Hub. This blog post however, explores the modernisation opportunities that you may not even have thought of yet, the business benefits, how to start preparing your organisation, empowering your teams, and identifying where GitHub Copilot can help. I’ve spent the last 8 months working with customers exploring usage of GitHub Copilot, and want to share what my team members and I have discovered in terms of new opportunities to modernise, transform your applications, bringing some fun back into those migrations! Let’s delve into how GitHub Copilot is helping teams update old systems, move processes to the cloud, and achieve results faster than ever before. Background: The Modernisation Challenge (Then vs Now) Modernising legacy software has always been hard. In the past, teams faced steep challenges: brittle codebases full of technical debt, outdated languages (think decades-old COBOL or VB6), sparse documentation, and original developers long gone. Integrating old systems with modern cloud services often requiring specialised skills that were in short supply – for example, check out this fantastic post from Arvi LiVigni (@arilivigni ) which talks about migrating from COBOL “the number of developers who can read and write COBOL isn’t what it used to be,” making those systems much harder to update". Common pain points included compatibility issues, data migrations, high costs, security vulnerabilities, and the constant risk that any change could break critical business functions. It’s no wonder many modernisation projects stalled or were “put off” due to their complexity and risk. So, what’s different now (circa 2025) compared to two years ago? In a word: Intelligent AI assistance. Tools like GitHub Copilot have emerged as AI pair programmers that dramatically lower the barriers to modernisation. Arvi’s post talks about how only a couple of years ago, developers had to comb through documentation and Stack Overflow for clues when deciphering old code or upgrading frameworks. Today, GitHub Copilot can act like an expert co-developer inside your IDE, ready to explain mysterious code, suggest updates, and even rewrite legacy code in modern languages. This means less time fighting old code and more time implementing improvements. As Arvi says “nine times out of 10 it gives me the right answer… That speed – and not having to break out of my flow – is really what’s so impactful.” In short, AI coding assistants have evolved from novel experiments to indispensable tools, reimagining how we approach software updates and cloud adoption. I’d also add from my own experience – the models we were using 12 months ago have already been superseded by far superior models with ability to ingest larger context and tackle even further complexity. It's easier to experiment, and fail, bringing more robust outcomes – with such speed to create those proof of concepts, experimentation and failing faster, this has also unlocked the ability to test out multiple hypothesis’ and get you to the most confident outcome in a much shorter space of time. Modernisation is easier now because AI reduces the heavy lifting. Instead of reading the 10,000-line legacy program alone, a developer can ask Copilot to explain what the code does or even propose a refactored version. Rather than manually researching how to replace an outdated library, they can get instant recommendations for modern equivalents. These advancements mean that tasks which once took weeks or months can now be done in days or hours – with more confidence and less drudgery - more fun! The following sections will dive into specific opportunities unlocked by GitHub Copilot across the modernisation journey which you may not even have thought of. Modernisation Opportunities Unlocked by Copilot Modernising an application isn’t just about updating code – it involves bringing everyone and everything up to speed with cloud-era practices. Below are several scenarios and how GitHub Copilot adds value, with the specific benefits highlighted: 1. AI-Assisted Legacy Code Refactoring and Upgrades Instant Code Comprehension: GitHub Copilot can explain complex legacy code in plain English, helping developers quickly understand decades-old logic without scouring scarce documentation. For example, you can highlight a cryptic COBOL or C++ function and ask Copilot to describe what it does – an invaluable first step before making any changes. This saves hours and reduces errors when starting a modernisation effort. Automated Refactoring Suggestions: The AI suggests modern replacements for outdated patterns and APIs, and can even translate code between languages. For instance, Copilot can help convert a COBOL program into JavaScript or C# by recognising equivalent constructs. It also uses transformation tools (like OpenRewrite for Java/.NET) to systematically apply code updates – e.g. replacing all legacy HTTP calls with a modern library in one sweep. Developers remain in control, but GitHub Copilot handles the tedious bulk edits. Bulk Code Upgrades with AI: GitHub Copilot’s App Modernisation capabilities can analyse an entire codebase and generate a detailed upgrade plan, then execute many of the code changes automatically. It can upgrade framework versions (say from .NET Framework 4.x to .NET 6, or Java 8 to Java 17) by applying known fix patterns and even fixing compilation errors after the upgrade. Teams can finally tackle those hundreds of thousand-line enterprise applications – a task that could take multiple years with GitHub Copilot handling the repetitive changes. Technical Debt Reduction: By cleaning up old code and enforcing modern best practices, GitHub Copilot helps chip away at years of technical debt. The modernised codebase is more maintainable and stable, which lowers the long-term risk hanging over critical business systems. Notably, the tool can even scan for known security vulnerabilities during refactoring as it updates your code. In short, each legacy component refreshed with GitHub Copilot comes out safer and easier to work on, instead of remaining a brittle black box. 2. Accelerating Cloud Migration and Azure Modernisation Guided Azure Migration Planning: GitHub Copilot can assess a legacy application’s cloud readiness and recommend target Azure services for each component. For instance, it might suggest migrating an on-premises database to Azure SQL, moving file storage to Azure Blob Storage, and converting background jobs to Azure Functions. This provides a clear blueprint to confidently move an app from servers to Azure PaaS. One-Click Cloud Transformations: GitHub Copilot comes with predefined migration tasksthat automate the code changes required for cloud adoption. With one click, you can have the AI apply dozens of modifications across your codebase. For example: File storage: Replace local file read/writes with Azure Blob Storage SDK calls. Email/Comms: Swap out SMTP email code for Azure Communication Services or SendGrid. Identity: Migrate authentication from Windows AD to Azure AD (Entra ID) libraries. Configuration: Remove hard-coded configurations and use Azure App Configuration or Key Vault for secrets. GitHub Copilot performs these transformations consistently, following best practices (like using connection strings from Azure settings). After applying the changes, it even fixes any compile errors automatically, so you’re not left with broken builds. What used to require reading countless Azure migration guides is now handled in minutes. Automated Validation & Deployment: Modernisation doesn’t stop at code changes. GitHub Copilot can also generate unit tests to validate that the application still behaves correctly after the migration. It helps ensure that your modernised, cloud-ready app passes all its checks before going live. When you’re ready to deploy, GitHub Copilot can produce the necessary Infrastructure-as-Code templates (e.g. Azure Resource Manager Bicep files or Terraform configs) and even set up CI/CD pipeline scripts for you. In other words, the AI can configure the Azure environment and deployment process end-to-end. This dramatically reduces manual effort and error, getting your app to the cloud faster and with greater confidence. Integrations: GitHub Copilot also helps tackle larger migration scenarios that were previously considered too complex. For example, many enterprises want to retire expensive proprietary integration platforms like MuleSoft or Apigee and use Azure-native services instead, but rewriting hundreds of integration workflows was daunting. Now, GitHub Copilot can assist in translating those workflows: for instance, converting an Apigee API proxy into an Azure API Management policy, or a MuleSoft integration into an Azure Logic App. Multi-Cloud Migrations: if you plan to consolidate from other clouds into Azure, GitHub Copilot can suggest equivalent Azure services and SDK calls to replace AWS or GCP-specific code. These AI-assisted conversions significantly cut down the time needed to reimplement functionality on Azure. The business impact can be substantial. By lowering the effort of such migrations, GitHub Copilot makes it feasible to pursue opportunities that deliver big cost savings and simplification. 3. Boosting Developer Productivity and Quality Instant Unit Tests (TDD Made Easy): Writing tests for old code can be tedious, but GitHub Copilot can generate unit test cases on the fly. Developers can highlight an existing function and ask Copilot to create tests; it will produce meaningful test methods covering typical and edge scenarios. This makes it practical to apply test-driven development practices even to legacy systems – you can quickly build a safety net of tests before refactoring. By catching bugs early through these AI-generated tests, teams gain confidence to modernise code without breaking things. It essentially injects quality into the process from the start, which is crucial for successful modernisation. DevOps Automation: GitHub Copilot helps modernise your build and deployment process as well. It can draft CI/CD pipeline configurations, Dockerfiles, Kubernetes manifests, and other DevOps scripts by leveraging its knowledge of common patterns. For example, when setting up a GitHub Actions workflow to deploy your app, GitHub Copilot will autocomplete significant parts (like build steps, test runs, deployment jobs) based on the project structure. This not only saves time but also ensures best practices (proper caching, dependency installation, etc.) are followed by default. Microsoft even provides an extension where you can describe your Azure infrastructure needs in plain language and have GitHub Copilot generate the corresponding templates and pipeline YAML. By automating these pieces, teams can move to cloud-based, automated deployments much faster. Behaviour-Driven Development Support: Teams practicing BDD write human-readable scenarios (e.g. using Gherkin syntax) describing application behaviour. GitHub Copilot’s AI is adept at interpreting such descriptions and suggesting step definition code or test implementations to match. For instance, given a scenario “When a user with no items checks out, then an error message is shown,” GitHub Copilot can draft the code for that condition or the test steps required. This helps bridge the gap between non-technical specifications and actual code. It makes BDD more efficient and accessible, because even if team members aren’t strong coders, the AI can translate their intent into working code that developers can refine. Quality and Consistency: By using AI to handle boilerplate and repetitive tasks, developers can focus more on high-value improvements. GitHub Copilot’s suggestions are based on a vast corpus of code, which often means it surfaces well-structured, idiomatic patterns. Starting from these suggestions, developers are less likely to introduce errors or reinvent the wheel, which leads to more consistent code quality across the project. The AI also often reminds you of edge cases (for example, suggesting input validation or error handling code that might be missed), contributing to a more robust application. In practice, many teams find that adopting GitHub Copilot results in fewer bugs and quicker code reviews, as the code is cleaner on the first pass. It’s like having an extra set of eyes on every pull request, ensuring standards are met. Business Benefits of AI-Powered Modernisation Bringing together the technical advantages above, what’s the payoff for the business and stakeholders? Modernising with GitHub Copilot can yield multiple tangible and intangible benefits: Accelerated Time-to-Market: Modernisation projects that might have taken a year can potentially be completed in a few months, or an upgrade that took weeks can be done in days. This speed means you can deliver new features to customers sooner and respond faster to market changes. It also reduces downtime or disruption since migrations happen more swiftly. Cost Savings: By automating repetitive work and reducing the effort required from highly paid senior engineers, GitHub Copilot can trim development costs. Faster project completion also means lower overall project cost. Additionally, running modernised apps on cloud infrastructure (with updated code) often lowers operational costs due to more efficient resource usage and easier maintenance. There’s also an opportunity cost benefit: developers freed up by Copilot can work on other value-adding projects in parallel. Improved Quality & Reliability: GitHub Copilot’s contributions to testing, bug-fixing, and even security (like patching known vulnerabilities during upgrades) result in more robust applications. Modernised systems have fewer outages and security incidents than shaky legacy ones. Stakeholders will appreciate that with GitHub Copilot, modernisation doesn’t mean “trading one set of bugs for another” – instead, you can increase quality as you modernise (GitHub’s research noted higher code quality when using Copilot, as developers are less likely to introduce errors or skip tests). Business Agility: A modernised application (especially one refactored for cloud) is typically more scalable and adaptable. New integrations or features can be added much faster once the platform is up-to-date. GitHub Copilot helps clear the modernisation hurdle, after which the business can innovate on a solid, flexible foundation (for example, once a monolith is broken into microservices or moved to Azure PaaS, you can iterate on it much faster in the future). AI-assisted modernisation thus unlocks future opportunities (like easier expansion, integrations, AI features, etc.) that were impractical on the legacy stack. Employee Satisfaction and Innovation: Developer happiness is a subtle but important benefit. When tedious work is handled by AI, developers can spend more time on creative tasks – designing new features, improving user experience, exploring new technologies. This can foster a culture of innovation. Moreover, being seen as a company that leverages modern tools (like AI Copilot) helps attract and retain top tech talent. Teams that successfully modernise critical systems with Copilot will gain confidence to tackle other ambitious projects, creating a positive feedback loop of improvement. To sum up, GitHub Copilot acts as a force-multiplier for application modernisation. It enables organisations to do more with less: convert legacy “boat anchors” into modern, cloud-enabled assets rapidly, while improving quality and developer morale. This aligns IT goals with business goals – faster delivery, greater efficiency, and readiness for the future. Call to Action: Embrace the Future of Modernisation GitHub Copilot has proven to be a catalyst for transforming how we approach legacy systems and cloud adoption. If you’re excited about the possibilities, here are next steps and what to watch for: Start Experimenting: If you haven’t already, try GitHub Copilot on a sample of your code. Use Copilot or Copilot Chat to explain a piece of old code or generate a unit test. Seeing it in action on your own project can build confidence and spark ideas for where to apply it. Identify a Pilot Project: Look at your application portfolio for a candidate that’s ripe for modernisation – maybe a small legacy service that could be moved to Azure, or a module that needs a refactor. Use GitHub Copilot to assess and estimate the effort. Often, you’ll find tasks once deemed “too hard” might now be feasible. Early successes will help win support for larger initiatives. Stay Tuned for Our Upcoming Blog Series: This post is just the beginning. In forthcoming posts, we’ll dive deeper into: Setting Up Your Organisation for Copilot Adoption: Practical tips on preparing your enterprise environment – from licensing and security considerations to training programs. We’ll discuss best practices (like running internal awareness campaigns, defining success metrics, and creating Copilot champions in your teams) to ensure a smooth rollout. Empowering Your Colleagues: How to foster a culture that embraces AI assistance. This includes enabling continuous learning, sharing prompt techniques and knowledge bases, and addressing any scepticism. We’ll cover strategies to support developers in using Copilot effectively, so that everyone from new hires to veteran engineers can amplify their productivity. Identifying High-Impact Modernisation Areas: Guidance on spotting where GitHub Copilot can add the most value. We’ll look at different domains – code, cloud, tests, data – and how to evaluate opportunities (for example, using telemetry or feedback to find repetitive tasks suited for AI, or legacy components with high ROI if modernised). Engage and Share: As you start leveraging Copilot for modernisation, share your experiences and results. Success stories (even small wins like “GitHub Copilot helped reduce our code review times” or “we migrated a component to Azure in 1 sprint”) can build momentum within your organisation and the broader community. We invite you to discuss and ask questions in the comments or in our tech community forums. Take a look at the new App Modernisation Guidance—a comprehensive, step-by-step playbook designed to help organisations: Understand what to modernise and why Migrate and rebuild apps with AI-first design Continuously optimise with built-in governance and observability Modernisation is a journey, and AI is the new compass and Copilot to guide the way. By embracing tools like GitHub Copilot, you position your organisation to break through modernisation barriers that once seemed insurmountable. The result is not just updated software, but a more agile, cloud-ready business and a happier, more productive development team. Now is the time to take that step. Empower your team with Copilot, and unlock the full potential of your applications and your developers. Stay tuned for more insights in our next posts, and let’s modernise what’s possible together!Search Less, Build More: Inner Sourcing with GitHub Copilot and ADO MCP Server
Developers burn cycles context‑switching: opening five repos to find a logging example, searching a wiki for a data masking rule, scrolling chat history for the latest pipeline pattern. Organisations that I speak to are often on the path of transformational platform engineering projects but always have the fear or doubt of "what if my engineers don't use these resources". While projects like Backstage still play a pivotal role in inner sourcing and discoverability I also empathise with developers who would argue "How would I even know in the first place, which modules have or haven't been created for reuse". In this blog we explore how we can ensure organisational standards and developer satisfaction without any heavy lifting on either side, no custom model training, no rewriting or relocating of repositories and no stagnant local data. Using GitHub Copilot + Azure DevOps MCP server (with the free `code_search` extension) we turn the IDE into an organizational knowledge interface. Instead of guessing or re‑implementing, engineers can start scaffolding projects or solving issues as they would normally (hopefully using Copilot) and without extra prompting. GitHub Copilot can lean into organisational standards and ensure recommendations are made with code snippets directly generated from existing examples. What Is the Azure DevOps MCP Server + code_search Extension? MCP (Model Context Protocol) is an open standard that lets agents (like GitHub Copilot) pull in structured, on-demand context from external systems. MCP servers contain natural language explanations of the tools that the agent can utilise allowing dynamic decision making of when to implement certain toolsets over others. The Azure DevOps MCP Server is the ADO Product Team's implementation of that standard. It exposes your ADO environment in a way Copilot can consume. Out of the box it gives you access to: Projects – list and navigate across projects in your organization. Repositories – browse repos, branches, and files. Work items – surface user stories, bugs, or acceptance criteria. Wiki's – pull policies, standards, and documentation. This means Copilot can ground its answers in live ADO content, instead of hallucinating or relying only on what’s in the current editor window. The ADO server runs locally from your own machine to ensure that all sensitive project information remains within your secure network boundary. This also means that existing permissions on ADO objects such as Projects or Repositories are respected. Wiki search tooling available out of the box with ADO MCP server is very useful however if I am honest I have seen these wiki's go unused with documentation being stored elsewhere either inside the repository or in a project management tool. This means any tool that needs to implement code requires the ability to accurately search the code stored in the repositories themself. That is where the code_search extension enablement in ADO is so important. Most organisations have this enabled already however it is worth noting that this pre-requisite is the real unlock of cross-repo search. This allows for Copilot to: Query for symbols, snippets, or keywords across all repos. Retrieve usage examples from code, not just docs. Locate standards (like logging wrappers or retry policies) wherever they live. Back every recommendation with specific source lines. In short: MCP connects Copilot to Azure DevOps. code_search makes that connection powerful by turning it into a discovery engine. What is the relevance of Copilot Instructions? One of the less obvious but most powerful features of GitHub Copilot is its ability to follow instructions files. Copilot automatically looks for these files and uses them as a “playbook” for how it should behave. There are different types of instructions you can provide: Organisational instructions – apply across your entire workspace, regardless of which repo you’re in. Repo-specific instructions – scoped to a particular repository, useful when one project has unique standards or patterns. Personal instructions – smaller overrides layered on top of global rules when a local exception applies. (Stored in .github/copilot-instructions.md) In this solution, I’m using a single personal instructions file. It tells Copilot: When to search (e.g., always query repos and wikis before answering a standards question). Where to look (Azure DevOps repos, wikis, and with code_search, the code itself). How to answer (responses must cite the repo/file/line or wiki page; if no source is found, say so). How to resolve conflicts (prefer dated wiki entries over older README fragments). As a small example, a section of a Copilot instruction file could look like this: # GitHub Copilot Instructions for Azure DevOps MCP Integration This project uses Azure DevOps with MCP server integration to provide organizational context awareness. Always check to see if the Azure DevOps MCP server has a tool relevant to the user's request. ## Core Principles ### 1. Azure DevOps Integration - **Always prioritize Azure DevOps MCP tools** when users ask about: - Work items, stories, bugs, tasks - Pull requests and code reviews - Build pipelines and deployments - Repository operations and branch management - Wiki pages and documentation - Test plans and test cases - Project and team information ### 2. Organizational Context Awareness - Before suggesting solutions, **check existing organizational patterns** by: - Searching code across repositories for similar implementations - Referencing established coding standards and frameworks - Looking for existing shared libraries and utilities - Checking architectural decision records (ADRs) in wikis ### 3. Cross-Repository Intelligence - When providing code suggestions: - **Search for existing patterns** in other repositories first - **Reference shared libraries** and common utilities - **Maintain consistency** with organizational standards - **Suggest reusable components** when appropriate ## Tool Usage Guidelines ### Work Items and Project Management When users mention bugs, features, tasks, or project planning: ``` ✅ Use: wit_my_work_items, wit_create_work_item, wit_update_work_item ✅ Use: wit_list_backlogs, wit_get_work_items_for_iteration ✅ Use: work_list_team_iterations, core_list_projects The result... To test this I created 3 ADO Projects each with between 1-2 repositories. The repositories were light with only ReadMe's inside containing descriptions of the "repo" and some code snippets examples for usage. I have then created a brand-new workspace with no context apart from a Copilot instructions document (which could be part of a repo scaffold or organisation wide) which tells Copilot to search code and the wikis across all ADO projects in my demo environment. It returns guidance and standards from all available repo's and starts to use it to formulate its response. In the screenshot I have highlighted some key parts with red boxes. The first being a section of the readme that Copilot has identified in its response, that part also highlighted within CoPilot chat response. I have highlighted the rather generic prompt I used to get this response at the bottom of that window too. Above I have highlighted Copilot using the MCP server tooling searching through projects, repo's and code. Finally the largest box highlights the instructions given to Copilot on how to search and how easily these could be optimised or changed depending on the requirements and organisational coding standards. How did I implement this? Implementation is actually incredibly simple. As mentioned I created multiple projects and repositories within my ADO Organisation in order to test cross-project & cross-repo discovery. I then did the following: Enable code_search - in your Azure DevOps organization (Marketplace → install extension). Login to Azure - Use the AZ CLI to authenticate to Azure with "az login". Create vscode/mcp.json file - Snippet is provided below, the organisation name should be changed to your organisations name. Start and enable your MCP server - In the mcp.json file you should see a "Start" button. Using the snippet below you will be prompted to add your organisation name. Ensure your Copilot agent has access to the server under "tools" too. View this setup guide for full setup instructions (azure-devops-mcp/docs/GETTINGSTARTED.md at main · microsoft/azure-devops-mcp) Create a Copilot Instructions file - with a search-first directive. I have inserted the full version used in this demo at the bottom of the article. Experiment with Prompts – Start generic (“How do we secure APIs?”). Review the output and tools used and then tailor your instructions. Considerations While this is a great approach I do still have some considerations when going to production. Latency - Using MCP tooling on every request will add some latency to developer requests. We can look at optimizing usage through copilot instructions to better identify when Copilot should or shouldn't use the ADO MCP server. Complex Projects and Repositories - While I have demonstrated cross project and cross repository retrieval my demo environment does not truly simulate an enterprise ADO environment. Performance should be tested and closely monitored as organisational complexity increases. Public Preview - The ADO MCP server is moving quickly but is currently still public preview. We have demonstrated in this article how quickly we can make our Azure DevOps content discoverable. While their are considerations moving forward this showcases a direction towards agentic inner sourcing. Feel free to comment below how you think this approach could be extended or augmented for other use cases! Resources MCP Server Config (/.vscode/mcp.json) { "inputs": [ { "id": "ado_org", "type": "promptString", "description": "Azure DevOps organization name (e.g. 'contoso')" } ], "servers": { "ado": { "type": "stdio", "command": "npx", "args": ["-y", "@azure-devops/mcp", "${input:ado_org}"] } } } Copilot Instructions (/.github/copilot-instructions.md) # GitHub Copilot Instructions for Azure DevOps MCP Integration This project uses Azure DevOps with MCP server integration to provide organizational context awareness. Always check to see if the Azure DevOps MCP server has a tool relevant to the user's request. ## Core Principles ### 1. Azure DevOps Integration - **Always prioritize Azure DevOps MCP tools** when users ask about: - Work items, stories, bugs, tasks - Pull requests and code reviews - Build pipelines and deployments - Repository operations and branch management - Wiki pages and documentation - Test plans and test cases - Project and team information ### 2. Organizational Context Awareness - Before suggesting solutions, **check existing organizational patterns** by: - Searching code across repositories for similar implementations - Referencing established coding standards and frameworks - Looking for existing shared libraries and utilities - Checking architectural decision records (ADRs) in wikis ### 3. Cross-Repository Intelligence - When providing code suggestions: - **Search for existing patterns** in other repositories first - **Reference shared libraries** and common utilities - **Maintain consistency** with organizational standards - **Suggest reusable components** when appropriate ## Tool Usage Guidelines ### Work Items and Project Management When users mention bugs, features, tasks, or project planning: ``` ✅ Use: wit_my_work_items, wit_create_work_item, wit_update_work_item ✅ Use: wit_list_backlogs, wit_get_work_items_for_iteration ✅ Use: work_list_team_iterations, core_list_projects ``` ### Code and Repository Operations When users ask about code, branches, or pull requests: ``` ✅ Use: repo_list_repos_by_project, repo_list_pull_requests_by_repo ✅ Use: repo_list_branches_by_repo, repo_search_commits ✅ Use: search_code for finding patterns across repositories ``` ### Documentation and Knowledge Sharing When users need documentation or want to create/update docs: ``` ✅ Use: wiki_list_wikis, wiki_get_page_content, wiki_create_or_update_page ✅ Use: search_wiki for finding existing documentation ``` ### Build and Deployment When users ask about builds, deployments, or CI/CD: ``` ✅ Use: pipelines_get_builds, pipelines_get_build_definitions ✅ Use: pipelines_run_pipeline, pipelines_get_build_status ``` ## Response Patterns ### 1. Discovery First Before providing solutions, always discover organizational context: ``` "Let me first check what patterns exist in your organization..." → Search code, check repositories, review existing work items ``` ### 2. Reference Organizational Standards When suggesting code or approaches: ``` "Based on patterns I found in your [RepositoryName] repository..." "Following your organization's standard approach seen in..." "This aligns with the pattern established in [TeamName]'s implementation..." ``` ### 3. Actionable Integration Always offer to create or update Azure DevOps artifacts: ``` "I can create a work item for this enhancement..." "Should I update the wiki page with this new pattern?" "Let me link this to the current iteration..." ``` ## Specific Scenarios ### New Feature Development 1. **Search existing repositories** for similar features 2. **Check architectural patterns** and shared libraries 3. **Review related work items** and planning documents 4. **Suggest implementation** based on organizational standards 5. **Offer to create work items** and documentation ### Bug Investigation 1. **Search for similar issues** across repositories and work items 2. **Check related builds** and recent changes 3. **Review test results** and failure patterns 4. **Provide solution** based on organizational practices 5. **Offer to create/update** bug work items and documentation ### Code Review and Standards 1. **Compare against organizational patterns** found in other repositories 2. **Reference coding standards** from wiki documentation 3. **Suggest improvements** based on established practices 4. **Check for reusable components** that could be leveraged ### Documentation Requests 1. **Search existing wikis** for related content 2. **Check for ADRs** and technical documentation 3. **Reference patterns** from similar projects 4. **Offer to create/update** wiki pages with findings ## Error Handling If Azure DevOps MCP tools are not available or fail: 1. **Inform the user** about the limitation 2. **Provide alternative approaches** using available information 3. **Suggest manual steps** for Azure DevOps integration 4. **Offer to help** with configuration if needed ## Best Practices ### Always Do: - ✅ Search organizational context before suggesting solutions - ✅ Reference existing patterns and standards - ✅ Offer to create/update Azure DevOps artifacts - ✅ Maintain consistency with organizational practices - ✅ Provide actionable next steps ### Never Do: - ❌ Suggest solutions without checking organizational context - ❌ Ignore existing patterns and implementations - ❌ Provide generic advice when specific organizational context is available - ❌ Forget to offer Azure DevOps integration opportunities --- **Remember: The goal is to provide intelligent, context-aware assistance that leverages the full organizational knowledge base available through Azure DevOps while maintaining development efficiency and consistency.**
