Why we invested
We led Vantic’s seed round three weeks after the first prototype existed, before evaluation had become a category anyone had named. The founding team had lived the problem inside a large research lab and knew precisely which failures mattered and which were noise.
The insight was unfashionable at the time: model quality is not a benchmark number, it is a distribution of behaviours against your own traffic. Everything Vantic built follows from taking that seriously — harnesses that run against production requests, regression suites owned by product teams rather than researchers, and a scoring layer that a non-specialist can read.
Today the product runs against some of the most demanding model deployments in the market, and the company has stayed deliberately small while revenue has not.
They committed on a Thursday after two conversations, at a stage when every other fund wanted another quarter of data. That decisiveness bought us six months.
Our involvement
- Round led
- Seed, $4.5M
- Time to term sheet
- 9 days
- Follow-on
- Series A, 2025