Most GTM tools use a foundational model API and call it a day.

We use LLMs for most things too, but sometimes it is like hitting a nail with a sword.

To make GTM intelligence scalable and fast, we now train some of our own models on terabytes of people and company data. Creating inferred attributes from scratch - things no data vendor sells.

We use Baseten to power that infrastructure. Their team came by the office this past week and brought their inference engineering book plus a lot of good energy.

This is turning out to be one of the more important bets we have taken.