Session

Session

Breakout Track #316:30 – 16:55

Improving the Accuracy of an NL2SQL Agent Built with Azure Durable Functions

As AI agents become more widespread, use cases for analyzing data and querying databases in natural language are on the rise. However, simply handing SQL generation over to generative AI makes it far from easy to achieve the accuracy and safety you expect. In this session, using an NL2SQL Agent built with Azure Durable Functions as our subject, we will share the challenges we faced while working to improve accuracy and safety, along with the trial and error that went into solving them.

SPEAKER

Toshiyuki Hara

Toshiyuki Hara

JAZUG (Japan Azure User Group) / Microsoft MVP for Microsoft Azure / NEXTSCAPE Inc.

Cloud Solution Architect