Designing the data layer for AI apps and agents with Azure Cosmos DB | BRK367
In this Microsoft Ignite 2026 session, Nick Blake, Loveleen Rekhi, Madeline Egan, and Gahl Levy discuss how operational data, retrieval, memory, and state fit together in an Azure Cosmos DB data layer.
Retrieval and agent grounding
The session uses real-world use cases and customer examples to explain data-layer design for AI applications and multi-agent systems. It covers retrieval patterns for grounding agents in current application data:
- Full-text retrieval
- Vector retrieval
- Hybrid retrieval
- Semantic retrieval
- Sparse retrieval
- Multi-vector retrieval
Conversation history, memory, and state
Beyond retrieval, the session examines how Azure Cosmos DB supports conversation history, memory, state, and personalization. These concerns are discussed alongside operational data and global production scale, connecting the information agents retrieve with the context applications need to maintain.
Session details and resources
- Event: Microsoft Ignite 2026
- Session: BRK367 | Data
- Format: Breakout
- Level: (300) Advanced
- Resource: Data next steps