

Building AI for high-stakes work
AI demos are easy. Trusting AI with work where mistakes cost real money, real cases, or real customers is harder. High Proof Discussions is a new event series from Tavrn and Speakeasy. Each one is a candid conversation with people building AI for work where getting it wrong isn't an option.
For the first one, we're asking a simple question:
What will it take to trust AI to do real work?
The conversation
Vitor Vavolizza, CTO, Tavrn
Sagar Batchu, CEO, Speakeasy
We'll talk about what makes AI reliable enough to hand real responsibility to: accuracy, observability, security, and keeping humans in the loop. We'll also cover what each team has learned shipping AI into high-stakes workflows, and where the line between "assist" and "do" sits today.
Schedule
6:00 PM: Doors open, drinks and bites
6:30 PM - 7:30 PM: Fireside chat and audience Q&A
7:30 PM: Networking
About the hosts
Tavrn builds AI that gets medical records from any US provider for law firms.
Speakeasy makes AI integration easy, helping companies connect their APIs to developers and AI agents.