1The Anatomy of a Frontier Lab: Research, Data, Infra, Product, SafetyFrontier labs win by tightening the loop between research, data, infrastructure, product, and safety, not by optimizing one team alone.Practitioner 2The Model Portfolio and the Router: Orchestration as the ProductWhen model capability becomes plentiful, the durable advantage is not one model but the system that chooses among many.Practitioner 3Build vs Buy: Vertical Integration and Dependency RiskThe build-or-buy question is really about which layers you must own, which you can rent, and which dependencies can break your business.Expert 4Specialization Economics: Distillation, Post-Training, and the Cost CurveA practical guide to when smaller specialized models beat larger generalists on quality, cost, latency, and operational fit.Expert 5Product-Grounded Training: RL Environments, Data Flywheels, and Eval InfrastructureFrontier capability compounds when training environments, rewards, product data, and evaluations are tied to real user work.Expert 6Where the Moat Really Is: Harness, Distribution, Data Rights, and GovernanceFrontier models matter, but durable advantage usually lives in the systems, rights, channels, and trust built around them.Expert