Build with Web IQ: From API call to cited answer | BRK313
Takahiro Tsuchiya, Brian King, and Deena Minwalla walk through the path from a Web IQ API call to a cited answer, with practical patterns for agents, copilots, and AI applications.
From retrieval to cited answers
Web information can provide fresh knowledge for AI applications, but grounding requires more than search. This session covers the end-to-end architecture for retrieving relevant information with Web IQ, generating answers grounded in that information, and connecting responses to supporting sources.
Optimizing from prototype to production
The presenters cover how three signals help teams evaluate and optimize the experience:
- Latency: How long the application takes to deliver a response.
- Token usage: The volume of tokens used during processing.
- Traces: Visibility into the application's execution flow.
The session focuses on practical patterns for carrying these considerations from an initial prototype into production.
Session details and resource
- Event: Microsoft Ignite 2026
- Session: BRK313 — Microsoft IQ
- Format and level: Breakout, 300 — Advanced
- Resource: Microsoft IQ next steps