B2B Automation Spotlight

The episode contains solid, actionable frameworks around AI governance (five controls, three pillars, four governance questions) and the distinction between deterministic vs. non-deterministic systems. However, much of the content rehashes familiar compliance playbooks (SOC2, ISO27001, PCI) and includes significant filler from the panel Q&A section. The core insights cluster in Trevor's presentation; the panel discussion adds modest incremental value.

“AI failures really start with governance, not necessarily with technology”

“we have a new paradigm where decisions aren't being shaped and the decisions themselves are essentially non deterministic”

The framing of AI governance through ownership, visibility, and evidence is competent but well-trodden. The non-deterministic nature of LLMs as a risk paradigm shift is the closest to novel, but Trevor himself treats it as an emerging observation rather than original research. Most other frameworks (risk tiering, humans-in-the-loop, data classification) are standard practice in regulated industries. Panel discussion recycles generic startup advice.

“if we put in one plus one today it may equal two to a system, an AI system, but tomorrow it could be 2.5”

“the speed of change is amazing”

Trevor Horwitz is a credible CISO and founder with 20+ years in cybersecurity and compliance - a genuine practitioner with skin in the game through TrustNet. The panel includes other operators (Mike McNeil from Fleet, Matt Brown from Weré, Joanna Henkel from Toshiba). However, Adnan Masoud's role is unclear ('PhD and chief AI architect') and he comes across as more thought-leader than operator. Trevor is the strongest voice; panel weakens overall caliber through inclusion of less-proven speakers.

“I've been involved in cybersecurity, risk management and compliance for over 20 years”

“I'm the founder and CISO for trustnet”

Lacks concrete numbers, timelines, or named case studies. Trevor discusses frameworks abstractly (five controls, three pillars) without example companies, specific incidents, or metrics. Panel Q&A references vague scenarios (manufacturers, retailers, food/cosmetics) but no dollar figures, revenue impact, or measurable outcomes. The lawyer hallucination anecdote is mentioned but not detailed. Heavy reliance on principles over evidence.

“the classic failures of a lawyer who used an AI to generate an argument and the AI essentially hallucinated and made up case law”

“I've had uh, definitely had uh, successes and had failures”

Host Michael Bernsweig asks competent setup questions but rarely pushes back or follows up with depth. Panel questions are surface-level and audience-generated rather than challenging. When Trevor claims AI will 'enable you to go faster' via controls, no one probes the trade-off or asks for evidence. The moderator doesn't press guests on contradictions (e.g., Trevor's determinism argument vs. Adnan's frontier capability edge). Conversational flow is polite but lacks intellectual friction.

“Fantastic. Um, Trevor, the next question came in for you”

“Great question, uh, I would start with the users, ask them what to uh, do some surveys”