AI Agents Seattle November

This month we are hosting one of our favorite speakers Jim Bennett (Arize AI) with a terrific talk on self improving agent loop. But that's not all, we are also stoked that our attendees Madhu Gurramkonda and Shipra Sonal (both from Amazon) got inspired by previous AI Agents meetups and are going to share their experience moving an analytics team from ad-hoc SQL to agentic AI!

Self-improving agentic loops by Jim Bennett

Developers are embracing AI coding tools, and AI code review tools. AI engineers are also embracing AI observability and evaluations to discover issues in their agents. Putting this all together, means the next logical step is self improving AI agents that can discover bugs and automatically fix them.

This session covers just that. We will take an AI app, add observability and evals to detect issues, then show how that can be connected up to a coding agent to build the full cycle, raising PRs to fix these issues as they are discovered.

Bio

Jim is the worlds most energetic dev rel and a highly-experienced AI engineer. He’s British, so sounds way smarter than he actually is, and lives in the Pacific North West of the USA. In the past he’s lived in 4 continents working as a developer in the mobile, desktop, and scientific space. He's spoken at conferences and events all around the globe, organised meetup groups and communities, and written books on mobile development and IoT. He is currently a Microsoft MVP for AI and Developer Tools.

He also hates and is allergic to cats, but has a 13-year-old who loves cats, so he has 2 cats.

How to Move a Data Team to Agentic AI by Madhu Gurramkonda and Shipra Sonal

Most data teams still answer questions one ad-hoc SQL request at a time. This talk walks through how an analytics team can move from ad-hoc SQL to agentic AI, in the order that worked for us:

1. The platform in infrastructure as code, with a curated document beside every table, so agents and engineers read the same source of truth.
2. One command-line path to the data for people and agents under scoped permissions, so recurring queries become reviewed files with one audit trail.
3. Runbooks turned into agent skills, reviewed like code.
4. An always-on agent last: an orchestrator that sends each question to read-only specialist agents, plus a validator that checks every claim against its live source.

We spend most of the time on the validator, and we'll share what broke along the way and how to tell whether your team is ready for each step. We'll also cover two patterns: agents that drive business analytics through multi-agent orchestration, and agents that speed up the data team's own engineering work (development, testing, deployment and on-call).

Bios:

Madhu Gurramkonda leads data analytics and AI at Amazon, with a background in program management and data and BI engineering. He builds AI agents that business teams use for everyday analytics.

Shipra Sonal is a Data Engineering Manager at Amazon in Seattle with nine years across data engineering, business intelligence, machine learning and generative AI. She has helped move an analytics team from using AI as a chatbot to running agents, built around a validator that checks every answer against its live source before anyone sees it. She serves on the program committee of the HBDA-LLM workshop at IEEE BigData 2026.