Speaker:: Gadi Evron Title:: Opening Words - "Research conferences aren't effective." Duration:: 11 min Video:: https://www.youtube.com/watch?v=uH-UggR2Txg ## Key Thesis Research conferences have a structural networking problem that leaves attendees isolated despite being surrounded by relevant peers. Evron argues we're at a uniquely pivotal moment in history where the shift from deterministic to nondeterministic computing demands that security professionals simply try to keep up — and that effort alone is enough to stay relevant. ## Synopsis Evron opens by noting that Unprompted received nearly 500 submitted talks while planning for 70–120, and that the event grew from a 120–200 person concept to over 800 attendees online plus the in-person crowd at Salesforce. He credits the community-driven energy for that growth and acknowledges being "out quite a little bit of money" to make it happen. He frames four philosophical points: (1) we are moving from deterministic to nondeterministic configuration, a phrase he credits to his co-founder Sunilu, which he finds clarifying for thinking about what security looks like now; (2) the startup playbook is broken in the AI era — three-month product-market-fit cycles no longer apply; (3) teams at every level feel "outmoded," but the remedy is simply trying; and (4) he closes by reiterating that message — you are not outmoded, English and effort are sufficient, and even CEOs need to be in tools like Claude Code or Cursor daily. He then recaps a presentation by Joe Stewart originally given at ACOD (another conference Evron organized) titled "researcher conferences suck." Stewart's core complaint: too many people, bad name tags, noisy bars, no mechanism to find people doing similar work. The proposed fix was a "baseball card" slide deck where each attendee posts a picture, LinkedIn, what they can offer, and what they need — allowing targeted pre-conference networking. Evron re-implements this for Unprompted via an email-based contact form. Evron closes with housekeeping: Slack threads for every talk (slide decks to be posted there), an unconference slot voted on via sticky notes, 20-minute talk slots, and a 30-second question rule with a Slack workshop channel for attendees who want to refine questions before asking. ## Key Takeaways - The shift from deterministic to nondeterministic computing is the defining frame for security in 2026 - Conferences fail at networking by design; the baseball-card format (photo, what you offer, what you need) is a structural fix - "Just try" is the actual answer to feeling outmoded by AI — you don't need to be at the state-of-the-art, just above the floor - CEOs who aren't personally using coding agents right now are putting their companies at risk - This is a special moment in time; the community gathered at Unprompted may not exist in the same form next year ## Notable Quotes / Data Points - ~500 talk submissions, planned capacity 70–120, final attendance 800+ including online - "If instead of investing in so many foundation labs and models, we just tried to teach people how to use these tools, we would be everywhere right now" - "If the CEO is not spending time on Claude Code... I don't believe that company will survive" - Three speakers and a key staff member were stranded in Israel due to closed airspace on conference day #unprompted #claude