Summary In today's episode, I explain how to enforce data boundaries and prevent AI agents from executing destructive actions with compromised credentials in the enterprise. Here's what this means for you. You treat AI agents exactly like untrusted human contractors and apply the same containment, access control, and oversight strategies you already use for people…
Pro tip: telling AI not to fabricate or hallucinate is unhelpful if you don’t tell it how. Here’s what someone wrote on LinkedIn recently as part of an “ultimate prompt hack” or whatever superlative they used. Never fabricate: – statistics – articles – people – URLs – etc. I shook my head. This will work…
Summary In today's episode, I tackle the enterprise-level question of how to avoid replacing vendor lock-in with agentic platform lock-in by building local AI inference capabilities. Here's what this means for you. You'll gain pricing independence and operational control over your AI by owning inference hardware rather than depending solely on cloud providers like OpenAI…
Luxury. Enterprise. Upmarket. If you’ve not been following markets – especially shadow markets – you have not seen these shifts in the last 3 months: Every AI company is chasing enterprise customers hard now. Anthropic’s swimming upstream as fast as it can, with its competitors on its heels. Consumer brands of every kind have pivoted…
Summary In today's episode, I walk through how to implement real governance when AI systems are the ones deciding when to escalate to a human. Here's what this means for you. Treating AI agents like junior employees with managers, reporting mechanisms, and clear performance metrics gives you a practical framework that avoids the governance illusion…
All language is code, and writing is coding. Yesterday during the SmarterX AI Writer’s Summit, one of my session attendees said, after hearing me speak about using agentic AI to write a fiction novel: ‘But am I “writing” a trashy romance novel or am I “programming” one.’ ALL language is code, and writing is coding.…
Summary In today's episode, I break down what a HIPAA compliant LLM actually means in practice for clinical enterprise deployments. Here's what this means for you. You must guarantee that patient data never leaves your control, which means running local AI infrastructure instead of leaning on cloud tools like ChatGPT or Claude. You'll also learn…
Summary In today's episode, I walk through what LLMs.txt is, why Google now considers it part of site health, and how to build one using real analytics data. Here's what this means for you. You learn that LLMs.txt does not boost AI visibility but serves as a curated briefing document that helps AI agents navigate…
Your company will probably be buying AI hardware at some point if you’re heavy users of it. I ran an assessment this morning in Opcode to see what Trust Insights Claude Code usage was for the last 30 days and see what we would have paid if we were on Enterprise instead of Max 20.…
