If Your AI Agent Goes Wrong, Whose Name Is On It?
Last year, Harvey Nash’s Digital Leadership Report found a third of firms seeing a return on AI. This year it says two-thirds. David thinks that’s rubbish, and Faye Murray, Field Chief Data Officer for EMEA at Dataiku, largely agrees: much of that ‘return’ is personal productivity, while agentic AI at scale is still struggling. The pattern she sees is pilots that start with the technology rather than the problem, with no baseline, no accountable owner and costs that never surface in the proof of concept. Meanwhile, employees are building agents faster than IT can track them, and bans only push the risk underground. We unpack governed self-service, governance proportionate to risk, why sovereignty means being able to switch providers without rebuilding everything, and the question every agent owner should answer before launch: is the action reversible? Press play before you sign off your next pilot. On today’s show David and Sam discuss a post from Megan Caywood, take a look at here ideas here. Tech Talks is powered by Harvey Nash, the leading global provider of technology and talent solutions. Click here to learn more about our services. The music on Tech Talks is created by London artist Lemzi.
