Behind the Agent: How Hazel Built an AI Coworker for Consumer Brands

​Hazel started out predicting customer lifetime value. Then its founders saw a bigger opportunity: an AI coworker that can answer any question a consumer brand has about its business.

To make that work, they had to build the data stack underneath it too.

​Patrik Devlin, Hazel's co-founder and CTO, joins host Alex Monahan to talk through how it's built. Bring your questions, and we'll take them live as we dig into:

• ​A separate database for every customer, on MotherDuck

• ​Context built from dbt documentation, so the agent understands the data

• ​Memory that learns from your questions over time