Multi-Turn Conversations in Microsoft Agent Framework (Part 3)

Welcome to Part 3 of my Microsoft Agent Framework series! 📦 Resources & Links: 🔗 Code: https://github.com/Sandesh-hase/Microsoft-Agent-Framework.git End to End Project: https://youtu.be/BAdcNKf5d2g Part1: https://youtu.be/Gxh6fef4jJU Part2: https://youtu.be/i_wfwkgaFpI In this video, we dive deep into multi-turn conversations and explore how the framework intelligently manages thread-level memory. Using two parallel threads — Product 1 and Product 2 — you’ll see how the agent maintains completely isolated context for each product while answering follow-up questions with full accuracy. This demo clearly showcases one of the most powerful features of the Agent Framework: 🔹 Thread-specific memory retention 🔹 Context isolation between conversations 🔹 Reliable multi-turn interaction based on previous user prompts 🔹 Realistic invoice-based product questioning 🔹 Fully asynchronous execution with Azure OpenAI You’ll also see how the agent correctly recalls previous product details when asked inside the same thread, while responding with no context available in a different thread — a perfect demonstration of memory isolation in multi-turn AI systems. If you're building real-world AI applications, chatbots, workflow agents, or customer-support systems using Azure OpenAI or the Microsoft Agent Framework, this episode will give you a practical understanding of how multi-turn logic works behind the scenes. Don’t forget to check out Part 1 and Part 2 of the series if you’re new here — and stay tuned for upcoming episodes! #MicrosoftAgentFramework #AzureOpenAI #MultiTurnConversation #AIChatbots #LLMApplications #AzureAI #OpenAI #AgenticAI #PythonAI #InvoiceAI #ThreadMemory #AzureDevelopers #AIEngineering