Skip to content

Details

Abstract
Data agents are most useful when they understand the data they work with, not only its structure. This talk explores how a context layer helps an agent make better decisions and write accurate SQL that stays grounded in the business. We will look at the building blocks of context, including ontology and a context graph, and how they come together so an agent can retrieve what is relevant, reason over it, check its own work and respond with confidence. We will also cover how to measure the incremental changes with evaluations. Attendees will leave with a clear mental model and practical patterns for building context-aware data agents.
A 10-minute live demo of a simple Data Agent using Langchain’s Deep Agent, follows the talk.
Details

  • Format: 60 minutes: a 35 to 40 minute talk, a 10-minute live demo and 10 minutes for Q&A.
  • Audience: data, analytics and AI engineers building agents on top of databases.
  • Level: Intermediate.

Bio
Saket is a Principal Data Engineer

He specializes in building modern and highly scalable data platforms. This also includes driving Data platform AI strategy and Governance.

Date & Time
Nov 13th 8:30 pm EST

Related topics

AI/ML
New Technology
Computer Programming
Open Source
Software Development

You may also like