How do we give AI agents enough autonomy to be useful while keeping security, visibility, and control?
An intimate, off-the-record dinner bringing together CISOs, security leaders, AI executives, and founders building and securing the next generation of agentic systems.
Backed by 
6:30–9:30 PM PST
A restaurant in Dogpatch, San Francisco
What changes when agents call tools, execute code, touch sensitive data, and act across production systems?
Authentication, least privilege, delegation, and guardrails when the user is an autonomous agent.
Defending against prompt injection, data leakage, compromised tools, and malicious behavior at runtime.
HOSTS, ADVISOR & ORGANIZER
The founders of Armorer Labs, with their advisor Paul Vixie.
Organized with AI Socratic.
HostArmorer Labs
Co-founder & CEO
In security since 2012. At AWS he led the development and security of the first customer-facing AI security agent, built red-team agents, and created the open-source scanner ASH. Black Hat USA 2026 speaker.
HostArmorer Labs
Co-founder & CTO
Led AI security science at AWS, building and evaluating AI systems for security. Co-author of ThreatForest and SIR-Bench, and a Black Hat USA 2026 speaker. Leads research, evaluation, and engineering at Armorer.
AdvisorArmorer Labs
Advisor, Armorer Labs · Director, SIE Europe U.G.
Director at SIE Europe U.G., previously VP, Distinguished Engineer and Deputy CISO at AWS Security (2022-2026). Founder of Farsight Security, ISC, MAPS (the first anti-spam company) and PAIX (the first neutral Internet exchange). Inducted into the Internet Hall of Fame in 2014 for his work on DNS; author of BIND and Cron and of many DNS and DNSSEC standards. Ph.D. in Computer Science, Keio University.
OrganizerAI Socratic
Founder, AI Socratic
Founder of AI Socratic, a community of AI engineers, researchers, and founders built around Socratic dialogues on the latest in AI. Organizer of the dinner.
Photo of Paul Vixie: ImaginingtheInternet, CC BY 3.0. Resized and converted to WebP.
OFF THE RECORD
Candid, technical, and laid-back.
In the spirit of AI Socratic.
Why now
As agents gain access to internal data, APIs, codebases, tools, and critical systems, they introduce a fundamentally new security surface. Armorer Labs builds an autonomous security engineer for AI agents: it finds the attack, fixes it, and proves the agent still does its job.
On the table
Socratic dialogues: everyone shares what they’re seeing, challenges assumptions, and asks questions. Nothing said at the table is attributed.
THE EVENING
Leaders across AI and cybersecurity, working out what the security stack for the agentic enterprise needs to become. Great people, food, drinks, and laid-back conversations.
Dogpatch, SF · Monday, November 16Drinks, snacks, and first conversations as guests arrive.
One table, one conversation, and some shared fun.
The table opens up: follow the threads that matter to you, connect with whoever you click with, and keep the talk laid-back yet high-signal.
Stay for a last drink and keep the conversation going.
ONE TABLE
Paul VixieAdvisorOpen seatRequestOpen seatRequestCristian LeoHostOpen seatRequestOpen seatRequestOpen seatRequestFederico UlfoOrganizerOpen seatRequestOpen seatRequestDaniel BegimherHostThe venue
The exact location is shared once your invitation is approved.
GET A SEAT AT THE TABLE
For CISOs, security leaders, AI executives, and founders working on agentic systems. Seats are limited and every guest is approved to keep the conversation candid and high signal.
Hosted by Daniel Begimher, Cristian Leo, Federico Ulfo
