Closing Keynote: Azure Cosmos DB as the Backbone for AI-Enabled Enterprise Applications
summary
H&R Block modernizes a cloud‑native tax platform using Azure Cosmos DB as the single storage layer for microservices and AI, claiming fast developer productivity and global replication for seasonal peaks. A real‑time analytics agent is built on .NET/C#, Semantic Kernel and Azure OpenAI, using Cosmos vector search with diskN, cosine similarity, text‑embedding‑3‑large (3072 dims). Practical notes: session consistency plus client session IDs, scale‑down floor at 10% of peak, SQL API fails on ~20 GB docs unless compressed or using a different API, two tax seasons in production.
AI-written summary · about
chapters (12)
- 0:00Azure Cosmos DB journey
- 1:09Cloud-native goals and architecture
- 2:58Scalability and workload handling
- 4:23Lessons learned with Cosmos DB
- 6:51AI at H&R Block
- 9:15From search to AI chat
- 12:11Why Cosmos DB powers AI
- 13:35Demo introduction
- 16:08Analyzing chat with AI
- 21:50Demo technical deep dive
- 25:45Future possibilities
- 26:49Tax filing reminder
related talks
03h 30mAmericas Live Stream
27mKeynote
01h 05mKeynote: Building Applications with the Modern SQL Using AI and Microsoft Fabric
12mBuilding Generative AI Apps with Azure Cosmos DB
01mLearn How Azure Cosmos DB Has Become a Game Changer for
01mWelcome
30mAI: Empowering us all
27mIntegrating Azure Cosmos DB with the Azure Ecosystem