The best-run Marketing Operations teams are starting to think like engineering teams — and two of them will show you exactly what that looks like. Clay and Snowflake used AI to solve operational problems their existing stack couldn't handle efficiently and built production systems to do it. This session is a look inside both: real workflows, real results, built by the people who run them. 

Agenda:

Clay’s GTM Infrastructure as Code: Clay runs their own GTM motion on the product they build, treating it as an engineering problem. A technical walkthrough of the live system: unified data layer across contacts and accounts, function-based orchestrators and account agents on top, fully observable and CLI-buildable. A look inside a production system, not a demo.

Two AI Tools Built In-House · Snowflake

PolarReply — Campaign reply inboxes are a problem every MOps team has and almost no one owns well. At Snowflake, that's 850+ replies a day, spiking past 2,200. A commercial platform existed to handle it at $110K/year. The team built instead. PolarReply classifies every reply, pulls contact data from email signatures to enrich the CRM, and syncs unsubscribes directly back to Marketo. It has processed 211,000+ replies in production. The session covers the decision framework, the full build, and three failures that made it through review and shouldn't have.

SnowScrub — Snowflake’s very own list import tool that helps automate spreadsheet-based processing with an AI-powered app that uses fuzzy matching and SQL validation to clean and enrich lists before they ever touch your MAP. The result: faster turnaround, clean data, and no additional tool cost.

When: Thursday, September 10, 2026 · 2:00 – 4:00 PM PT

Venue: Snowflake HQ · 135 Constitution Drive, Menlo Park, CA 94025

In-person only. No replay. Seats are limited.

We'll be closing out the session with a raffle exclusively for MUG attendees. Don't miss your chance to win.