Agents You Can Verify - Infrastructure and standards for sovereign enterprise AI

​There is a buzz in the air around open AI standards vs frontier models for enterprises. Many enterprises are moving their most sensitive data into AI systems that span multiple models, tools, agents, and partners. “Sovereign AI” has come to mean more than where the model runs. It means knowing what informed an answer, who authorized the use, and what happened to the information afterward — and being able to prove it to a customer, a regulator, an insurer, or even a court. This evening brings together four people building the layers of that stack, from 2 frontier labs, one considered the leader in open-weight ecosystems & the other the agentic benchmark, an open-source agent platform, and a standards developer & secure data auditor. Each talk picks up where the last one leaves off:

Saurya Velagapudi — Principal Engineer, openhands.dev - presents:

Vasanth Chandra - Enterprise AI Products - Risk and Safety, META - presents:

Ben Curtis — CTO, JLINC & Protocol Developer - presents: "Establishing Trust By Not Trusting"

Bradley Taylor — AI Architect, Anthropic - presents: "A Receipt For Every AI Result"

The talks are followed by a panel with all four speakers and open Q&A.

Who should attend:
CISOs, CTOs, AI platform and infrastructure leads, start up founders building in this space, standards participants in town for IETF 127, and anyone responsible for putting agents into production with data they can’t afford to get wrong.

Schedule
6:00 — Doors, food and drinks
6:30 — Welcome and framing
6:35 — Four talks
7:30 — Panel and audience Q&A
8:00— Networking til close

Space is limited. Registration is required and will be confirmed.

​Wednesday, November 18, 2026 · 6–9 PM

San Francisco · held during IETF 127 week
Hosted by JLINC & Oasis Collective