#35 Scaling Agentic AI | Andy Eyherabide & Santiago Merchán, Wizeline

Summary We're joined by Andy Eyherabide, Chief AI Officer at Wizeline, and Santiago Merchán, AI Engineering Tech Lead at Wizeline, for a conversation about how established companies can compete as agentic AI reshapes software and business strategy. We unpack why strategy and execution are becoming separate disciplines now that AI can absorb the execution layer, why proprietary data alone isn't a moat without fast feedback loops, and how incumbents might borrow Peter Diamandis's "twin company" model to reinvent themselves without dragging along organizational baggage. Andy and Santiago make the case that governance, not technology, is the real bottleneck standing between the 23% of companies actually scaling agentic AI and everyone else stuck experimenting, and they lay out what a genuine AI harness and a real "sense, decide, act, check" policy loop look like in practice. Chapters 00:00 Why agentic AI is forcing every industry to rethink strategy 01:10 Separating strategy from execution — and what remains distinctly human 03:40 Is proprietary data really a moat? 06:19 The "twin company" model for reinventing an incumbent 07:51 Case study: how DirecTV spun up a streaming division to escape legacy resistance 10:03 Trust as the real differentiator over speed 11:02 Deciding which tasks an agent can run on its own 11:55 The shift from developer to product manager 13:13 Why the code can be a black box but the architecture can't 14:29 What separates companies that scale agentic AI from those stuck experimenting 16:28 Claude's dominance in coding models — and why "gold standard" won't last 18:20 Context management over prompt engineering 20:49 Trust but verify: staying accountable for AI-generated code 22:25 Why only 23% of companies are actually scaling agentic AI 23:50 Building an AI harness and a real governance policy loop -------------------------- -------------------------- Resources Andy Eyherabide - https://www.linkedin.com/in/aeyherabide/ Santiago Merchán - https://www.linkedin.com/in/santiagomerchan/ Sequoia Capital, "Services: The New Software" (intelligence vs. judgment framework) - https://sequoiacap.com/article/services-the-new-software/ McKinsey, "The state of AI in 2025" (source of the 23%-scaling-agentic-AI stat) - https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai Moonshots with Peter Diamandis (podcast) - https://www.diamandis.com/podcast -------------------------- -------------------------- -------------------------- -------------------------- -------------------------- --------------------------