No Bullsh!t Leadership

The episode contains a handful of genuinely useful ideas - the inward/outward leadership distinction, the talent pipeline decay at consulting firms, and the 10/80/10 task-split framework - but these are buried under repeated assertions that 'leadership hasn't changed since the Industrial Revolution' and prolonged program promotion. The ratio of novel ideas to filler and throat-clearing is poor for a 51-minute runtime.

“The drive to automate first without understanding the business process deeply enough is the reason 9 out of 10 AI projects fail”

“it's chewed out the bottom end of what they were doing in their business, which they had outsourced. They'd offshored that to the Philippines. So it's the offshoring piece that's going to go first”

The 'leading AI is inward, leading people is outward' framing is the one genuinely fresh conceptual contribution; everything else - automate last (cited from another author's book), AI as over-enthusiastic intern, change management as change management - is recycled or borrowed. The Industrial Revolution comparison is invoked so frequently it becomes a crutch rather than an insight.

“leading AI has an inward focus. Leading people has an outward Focus”

“AI is the ultimate people pleaser”

James Killick is a relevant practitioner - he built and sold a company, is currently building AI systems for SMBs, and speaks from operational experience rather than theory - but his credentials are described vaguely ('he built a company and then sold it') and his scale of work is unclear. He functions more as a knowledgeable peer than a marquee enterprise AI authority.

“James does this for a living. He built a company and then he sold it. And then he spent five years nearly running his next one into the ground”

“there's a concept where you use the different AI models against each other to work out whether uh, it's giving you good information. So if uh, you get it, Claude, you put it through chat GBT”

There are a handful of concrete anchors - the Philippines offshoring example, the $1 - 2K prototyping cost claim, and the rough McKinsey graduate-intake figure - but the 80 - 90% AI failure rate and most client outcomes are asserted without source or hard numbers, and the McKinsey estimate is immediately hedged into vagueness ('maybe a quarter of that, maybe less').

“Before it would be 50, $100,000 to implement something. Now you can do it for one or $2,000”

“Let's say a firm like McKinsey, let's say they used to take in 500 graduates a year out of the top universities and colleges in the US. Maybe now they're taking a quarter of that, maybe less”

Emma surfaces some structurally good questions - the talent pipeline problem, the ISO-standards gap, the inward/outward prompt - and Marty turns the tables with a direct question to James, which adds dynamism. However, the host allows the conversation to drift into extended car-analogy banter, permits repeated program promotion without redirecting, and never challenges the unsourced failure-rate statistics or pushes James on the actual scale of his client work.

“What can leaders do if they're not making the strategic decisions? They're not the ones setting the AI strategy. Maybe they've got directives coming down from the top. Their staff don't want to implement it”

“Are you thinking that this type of infrastructure can and will be built out in respect of AI or do you think it's moving too fast for that to catch up?”