The Velocity Room

Signal Series is a wrap. Thank you to everyone who spent the day with us at SPIN San Francisco! This was TVR’s first hosted event. We wanted a room where practitioners could talk honestly about running AI in production, and you made that happen. Thank you to Tom Edwards, Zane Niyazbekov, Christine Boles, Joseph Bisaccia, and Cal Al-Dhubaib for sharing what’s actually working and what isn’t. If you were in the room, what’s the one idea you’re taking back to your team?

Community AMA has started! #TVRsignal

An AI system that works in a demo isn’t the same as one you can trust in production. Cal Al-Dhubaib, Principal Technologist at Rubrik, broke down the gap

Joseph Bisaccia, Lead AI Engineer at Behavior Frontiers, walked the room through what it takes to build AI that holds up under HIPAA. If you’re deploying AI somewhere the rules are strict, this was the session to hear!

Christine Boles, MC Consulting Services, delivering a fireside chat illustrating how companies need to address edge compute and AI in real-world applications in the industrial and energy sector.

Brad Tompkins, Operations Lead, and Richard Piasentin, Board Member, leading a discussion on securing AI in production. What's your team's biggest challenge right now?

Zane Niyazbekov and Richard Piasentin leading the conversation on how hyperscale data centers actually get built.

Kicking the day off with Tom Edwards, EY, on the rise of agentic enterprise.

Doors are open at TVR Signal Series! Looking forward to a great day here in San Francisco.

Tomorrow in SF: a full day on scaling agentic AI, data centers, edge compute, and AI security. It ends with an AMA built from the room’s questions, then happy hour. TVR Signal Series, 9:45am PT at SPIN. Free. → hubs.li/Q04D0Xz50

2 days out. What’s the AI question your team can’t get a straight answer to? Bring it to TVR Signal Series SF on Thursday. You’ll submit questions at lunch, and the top ones close the day. Free and in person. Seats are limited, so RSVP to save yours: → hubs.li/Q04D0WD50

Three days until TVR Signal Series in San Francisco! Thursday covers the full AI stack: scaling agentic AI, building hyperscale data centers, physical AI at the edge, and keeping autonomous systems secure. Speakers include leaders from EY, AWS, and Rubrik. Free and in person at SPIN San Francisco. RSVP to save your seat: hubs.li/Q04D0VCs0

One week until TVR Signal Series SF. A day on scaling agentic AI, the infrastructure behind it, and keeping it secure, with speakers from EY, AWS, Rubrik, and Behavior Frontiers. Free, Nov 5. → hubs.li/Q04CD0Jj0

The best way to secure an AI agent: limit what it can reach. No crown jewels. No secrets. Segment it, baseline how it should behave, and contain it when it gets creative. How is your team deciding what your agents can reach?

Most event Q&As run out of time before the good questions come up. At TVR Signal Series, you rank the questions at lunch and the closing AMA starts at the top. SF, Nov 5. Free. → hubs.li/Q04CxG9-0

A lot of noise about “Black October.” The part worth your attention: two critical PHP flaws allow unauthenticated code execution on WordPress, Drupal, and Joomla. AI helped find them and build the exploit in under a week. Patched Oct 21. Check your PHP versions. hubs.li/Q04Cmvg90

2016: Microsoft’s Tay chatbot gets hijacked because nothing filtered what went in or out. ChatGPT had the same weak spot, but it was still just a mouth. No fingers. AI agents can take action now. What does that change about how your team thinks about these risks?

Two weeks out from TVR Signal Series in San Francisco. One day on scaling agentic AI, building data centers, physical AI at the edge, and keeping agents in check. Speakers from EY, AWS, Rubrik, and a 33-year Intel veteran. Free, Nov 5. → hubs.li/Q04BCtwX0

Hyperscale data centers start as a logistics problem. Land, power, water, steel, permits. Zane Niyazbekov (AWS) walks through how they get built at TVR Signal Series SF on Nov 5. RSVP and we'll save you a seat → hubs.ly/Q04BClth0

NIST, NSA, and CISA published ten practices for AI agents that act outside their limits. Voluntary, but a useful checklist: allowlist network access, keep logs the agent can’t touch, shut down automatically when a control fails. Which of the practices would you struggle to show evidence for in your environment right now? hubs.ly/Q04BChg00