Machine learning inside Snowflake, not next to it
This is Snowflake for machine learning: build a recommendation engine or a churn model with Cortex, using the data already sitting in your tables. No separate ML platform, no export step, and no new team to hire.
Jeff Springer, live.
Trains, runs, and infers directly in Snowflake, with no unload-retrain-drift cycle.
The old way: move the data out, then wait
Most ML platforms need the data to leave Snowflake first. It lands in a separate environment, gets modeled there, and then results need to get pushed back before anyone in the business can use them. Every step adds latency, a new governance boundary, and a new bill.
We hear this often from data leaders after they buy in to a new suite of platforms: the licenses arrive before the pipeline between them does, so data pipelines sit while a parallel ML build gets scoped and staffed. The churn model everyone wants stays a slide in a deck a little longer than anyone planned.
Build models with Snowflake Cortex, not a second platform
DAS42 builds recommendation engines, churn models, and anomaly detection directly inside Snowflake with Cortex. Training data, model code, and predictions stay in one place.
A churn model reads live subscriber or player data, scores accounts, and writes results back to the same tables your BI dashboards already query, without a duplicate copy of that data sitting in a third-party ML tool or a second pipeline to maintain.
See it in action
Teresa Kovich, full demo, live.
Shows genre/tag data quality work that directly feeds a recommendations engine, plus anomaly detection on media brand tweet volume.