Chris O’Brien is CTO of Advania UK and has been a Microsoft MVP for nineteen years. He has spent the last few years focused almost entirely on AI: building services around it, advising clients on it, and leading a large part of Advania’s own transformation. He is also, in his own words, pro AI and anti misuse. This episode is about the distance between those two things.

What Chris sees is adoption racing ahead while ownership of the output quietly falls away. People are producing pitch decks, vendor proposals, strategy documents and competitor analysis with AI in the driver’s seat, then shipping the results without the validation and fact checking that would normally accompany anything going out under an organisation’s name. The public examples are already piling up, including consultancy work delivered to a government client that turned out to contain hallucinations and had to be partly refunded.

What makes this an ownership conversation rather than a quality one is Chris’s test. Before anything leaves his hands he wants to stand behind every last bullet point, tagline and framing message, which means he removes as much as he adds. David Bowman picks that up as a form of subtraction and ties it back to the season’s theme: if you genuinely understand the domain, you can tell what does not need to be there.

The conversation also covers what vibe coding has and has not made easy, the failure modes that are genuinely dangerous rather than merely irritating, why token maxing and agent counts are the wrong measures of AI success, and whether it is currently career limiting to be the one voice in the room saying this will not work.