{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Leadership, Untangled","title":"When AI Makes a Five-Year Business Plan Impossible | Keith Mackay","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/88a460fe\"></iframe>","width":"100%","height":180,"duration":2929,"description":"Keith MacKay has spent 35 years inside the software economy as a CTO, founder, board advisor, and teacher, and now advises private equity and corporate clients as Managing Director of the Software Strategy Group at EY-Parthenon. In this conversation with Blaire, he breaks down what's actually happening when investors try to put a value on a software business in a world where AI capabilities are improving by roughly 40x a year, and a five-year hold period now spans a timeframe that's almost impossible to model with any confidence.\nThe conversation moves well beyond investment strategy into something every leader needs to think about. Keith makes the case that the people having the most success working with AI aren't necessarily the most technically gifted, they're the ones who can communicate clearly, break problems into pieces, and hold context across a conversation. He also gets into why AI adoption is tracking almost exactly the same curve as cloud computing, why treating AI as a cost-cutting tool rather than an augmentation tool tends to backfire, and what the coming squeeze on token economics might actually mean for how organisations make better decisions about where AI is really adding value.\nAI capabilities are improving at roughly 40x year on year, making five and seven-year investment horizons almost impossible to plan around with confidence\nOlder developers are outperforming younger ones with AI tools, because the skill that matters most is being able to clearly describe and break down a problem, which is a management skill, not a coding skill\nThe waterbed problem: accelerating code writing with AI just pushes the bottleneck into testing, deployment, and change management\n\"It's never technology, always psychology\" — accountability can't live inside an AI, and organisations that forget that create expensive problems downstream\nContext is everything: AI has no memory between sessions, and in long sessions the quality of responses degrades as earlier...","thumbnail_url":"https://img.transistorcdn.com/rYZpjaoLRn-N72MoIvwACKOTWAojg2s8ZQCIRsaM8Hc/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9mMDg4/MTE3Njc0MjMxY2Rh/MjgxYWZjODZjMGY3/ZDQ2Zi5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}