Brian Scanlan, Senior Principal Engineer at Intercom, the company building Fin.ai, joins Corey Quinn on Screaming in the Cloud to discuss Intercom’s move from AWS Aurora to PlanetScale’s managed Vitess after years of scaling challenges with their Ruby on Rails monolith. He explains how 13 Aurora clusters created operational pain and why PlanetScale’s white-glove, partnership-driven model won out over Amazon’s building-block approach.

The discussion also covers Intercom’s volunteer-based on-call system, their pivot to AI agents after ChatGPT’s launch, concerns about the shrinking pipeline of systems engineers, and how companies like PlanetScale and Snowflake are outpacing AWS by delivering superior user experiences.

About Brian: Brian is an engineer based in Intercom’s Dublin office. He fixes problems, builds things, and grows people.

Show Highlights

(01:34) The Digital Clippy Rant

(2:16) The Good Chatbot vs. Bad Chatbot 

(03:51) The AI Chatbot Revolution 

(04:33) Unexpected Consequences of Good Chatbots 

(05:42) AI Support vs. Human Support 

(05:59) The Alexa Problem and Feature Discoverability 

(19:03) Amazon's Struggles Moving Up the Stack 

(26:55) The Unix Networking Society Origins 

(34:43) The Global On-Call Challenge

(42:09) LinkedIn: The World's Largest Porn Site 

Links

Intercom: intercom.com

Brian on LinkedIn: https://www.linkedin.com/in/scanlanb/

Brian on Bluesky: https://bsky.app/profile/bscanlan.bsky.social

Sponsor
Wiz - Listen to Crying Out Cloud: wiz.io/crying-out-cloud