Why it matters
These founders chose to own their models and inference rather than call an API — a direct look at when self-hosting inference beats routing to a provider, and why.
The tokenmaxxing angle
Four builder case studies on the real make-vs-buy call for inference: what pushed each startup to own model-serving instead of paying per-token to a foundation-model API.
From the organizers
Hosted by Baseten and CapitalG at Baseten HQ, SF; 15-minute demo slots for 4 founders (Arjun Karanam, Jack O'Brien, Saad Jamal, Kenneth Acquah); described as space-limited.