AI Deep Dive (Virtual) - Building Self-Improving Agents

This is virtual event for our AI global community, please double-check your local time. Can't make it live? Register anyway to receive the webinar recording.

Join Snowflake to learn how to build self-improving agents

Tech Talk: Your Agent Should Fix Itself: Building Self-Improving Agents

Speaker: Elliot Botwick, Principal AI/ML Architect, Snowflake

Abstract: In this talk, we will share how coding agents help developers build high quality agents faster.

A key insight from building agents in production is that high quality agents operate with their goals, plans and actions aligned. We introduce the Agent Goal-Plan-Action (Agent GPA) framework to capture this insight, which achieved state of the art benchmarks on TRAIL/GAIA with 95% error coverage and 86% error localization.

The Agent GPA framework assesses the full agent's process:

• Was the goal achieved efficiently?
• Did the plan make sense?
• Were the right tools used?
• Did the agent follow through?

Without visibility into these steps, teams risk deploying agents that look reliable but create hidden costs in production. Inaccuracies can waste compute, inflate latency and lead to the wrong business decisions, all of which erode trust at scale.

We will show how to use coding agents to automate the process of measuring and improving your agent's GPA by using optimization skills that take advantage of the GPA evaluation framework. By the end, you’ll be able to use coding agents and the GPA framework to identify common agent failures, improve their agent and make it ready for production.

Venue:

Virtual, join from anywhere

More virtual sessions:

• July 22nd: AI Deep Dive with Google Ep1.
• July 29th: AI Deep Dive with Snowflake.
• Aug 5th: AI Deep Dive with Google Ep 2.
• Sep 2nd: AI Deep Dive with Google Ep 3.