Why Agent Scaffolding Beats Raw Frontier Models in Production
NVIDIA and CrowdStrike explain why custom agent architecture and proprietary telemetry beat generic frontier AI models in enterprise defense.
🔴 LIVE From CrowdStrike's Fal.Con 2026
With John Coogan, Jordi Hays, Jordan Curts, Michael Sentonas, Daniel Bernard, Justin Banano · Sunday, September 6, 2026
Broadcasting live from CrowdStrike's Fal.Con conference, John Coogan and Jordi Hays interview CrowdStrike leadership and NVIDIA executives about new AI-driven cybersecurity products, including SafeMind and Falcon Guardian. The discussions explore the rise of 'agent states,' AI attack economics, token budgeting hurdles for enterprises, and the role of specialized model harnesses in defense, followed by a breakdown of recent tech headlines including GoPro's acquisition and Apple's leadership trajectory.
Key takeaways
- NVIDIA research called AVO proved that wrapping a model in a dedicated reasoning and tool system lifted benchmark accuracy on ARC-AGI from 30% to 100%. Read more →
- AI agents inside company networks act like drunk interns with broad permissions, creating major security exposure. Read more →
- GoPro signed a definitive agreement to merge with privately held Starman Optical in a $285 million deal after creator Markiplier acquired an 8.5% stake. Read more →
- Model obliteration removes guardrails from open-weight models on platforms like Hugging Face, allowing anyone with a local machine to generate full ransomware packages. Read more →
- Enterprise token shock is real: half of companies discover unexpected monthly bills of roughly $20,000 more per employee than budgeted. Read more →
5 articles from this episode
Why Enterprise AI Agents Are Like Drunk Interns
CrowdStrike and NVIDIA launched SafeMind to defend enterprise networks against rogue AI agents. Here is how continuous agent defense works.
Apple's AI Trap, GoPro's $285M Exit, and the MrBeast Lottery
Why MrBeast needs a $1M cash lottery to sell books, GoPro's $285M merger, and how Apple's operational scale falters in the age of AI.
CrowdStrike on Why Open AI Models Create the 'Agent State'
CrowdStrike's Jordan Curts explains how obliterated AI models give lone hackers nation-state capabilities, shifting cyber defense to compute speed.
Why AI Phishing Click Rates Jumped from 12% to 60%
CrowdStrike President Michael Sentonas breaks down token shock and why AI phishing click-through rates spiked to over 60 percent.
More TBPN episodes
- $1M Trading Cards Heist, Trevor Milton Returns, The Death Of Common Courtesy | Diet TBPNOct 25 · 1 articles
- Jay Leno on Comedy, Supercars, and the State of HollywoodOct 25 · 2 articles
- 50 Gigawatts To The Face, 17% GDP Growth Please, Road to COD 4, Bing Gordon JoinsOct 25 · 6 articles
- SELL EVERYTHING (Including Your Dollars), PPI God Candle, Lee Jae Myung Says "We Must Live"Oct 18 · 2 articles