Designing Reliable Multi‑Agent Apps with Azure Cosmos DB | OD820

Author: Microsoft Developer - Bewertung: 0x - Views:4

AI agents are easy to prototype and hard to ship. Learn how to build a multi‑agent, multi‑tenant app that stays fast and reliable in production. See how Azure Cosmos DB powers agent memory, retrieval, and coordination using document processing, vector and full‑text search, and semantic reranking to deliver low‑latency responses at scale.



To learn more, please check out these resources:

* https://aka.ms/build26/OD820











𝗦𝗲𝘀𝘀𝗶𝗼𝗻 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻:

This is one of many sessions from the Microsoft Build 2026 event. View even more sessions on-demand and learn about Microsoft Build at https://build.microsoft.com



OD820 | English (US) | Cloud platform & data





Pre-recorded | (300) Advanced





#MSBuild



Chapters:

0:00 - Introduction and overview of multi-agent travel assistant using Azure Cosmos DB

00:02:18 - Importance of personalization and reliability in user experience

00:02:31 - Explanation of multi-agent pattern versus single-agent architecture

00:04:16 - Introduction to agentic memory and comparison with RAG

00:06:00 - Detailed architecture of the travel assistant multi-agent system

00:08:35 - Defining and implementing state, short-term, and long-term memory

00:11:06 - Four types of long-term memory: facts, procedural, episodic, and user summary

00:19:00 - Memory extraction, conflict resolution, and intelligent memory management demonstration

00:28:00 - Cosmos DB Agent Kit for best practices and developer productivity

00:34:50 - Reliability, active-active architecture, distributed transactions, integrated embeddings, and global secondary indexes with key takeaways