CIO Classified

The episode contains a handful of genuinely non-obvious ideas - humans degrading AI recommendations in aggregate, 'born in digital' decisions that couldn't exist without AI, and the deterministic/agentic hybrid distinction - but they are buried under lengthy career recaps, the relentlessly repeated self-driving car analogy, and platitude-heavy host framing that inflates runtime without adding substance.

“whenever the humans are touching the system and are messing with the recommendation they actually degrade the performance. Not all the time, but in aggregate it does”

“there is a whole new array of decisions that are born in digital that could not be resolved by people because the value of the decision is real time”

The Zoox-vs-Waymo framing for the enterprise stack collapse is a genuinely interesting riff, and the 'decision memory as institutional knowledge that survives employee churn' angle is fresh; but the self-driving car analogy is the company's own 2017 launch concept recycled throughout and the broader AI-agents-vs-copilots narrative is saturated territory by 2024 - 25 standards.

“a Waymo makes no sense. It's a fully equipped car for a driver to operate... And on top you stuck a bunch of raiders, lighters, sensors and intelligence to make it drive autonomously. And then you have the Zoox... you remove the human capability”

“the Autonomy is not another version or better version of my planning tool or my replenishment tool. It's not that it replaces the need to have a human touch that software”

Fred Laluyaux is a genuine serial operator - helped scale Anaplan to a Thoma Bravo acquisition and has run Aera for nine years deploying at Fortune 100 companies - but this is a sponsored episode where he is selling his own product, which structurally caps the candor and objectivity of insights.

“Left SAP, built Anaplan, uh, for a few years, did good success there”

“to this day I think ERA has digitized more than 50 million decisions for the likes of Exxon, Unilever, Estee”

There are credible, publicly attributable data points - Hershey's analyst-day slide projecting $50M savings and $100M inventory reduction, a German client harmonizing 42 ERP instances, named Fortune 100 customers, and the concrete 4-minute/4-hour/4-day deployment roadmap - though several compelling examples are anonymized ('a customer in New York,' 'one of our customers') and the Gartner 50%-by-2027 projection is cited without a source link.

“the Hershey Company... on their presentation a slide that talked about how Decision Intelligence was going to save the company $50 million and reduce inventory level by a million dollar by $100 million within the next couple of years”

“we have a client in Germany that harmonizes I think 42 instances of transactional systems into AI”

Yusuf Khan delivers multi-sentence, self-referential question preambles that eat runtime and are frequently more testimonial than interrogation; Ian Faison asks more grounded practical questions but neither host challenges extraordinary claims (e.g., that human intervention universally degrades performance), and the sponsor relationship visibly softens the entire exchange into a product showcase rather than a critical conversation.

“you've always driven that I want to make sure that people understand like it's not just oh, is a product et cetera”

“I, I. It's been quite the journey for you and I've learned a ton. Um, so we're going to get straight into it”