Episode 121 of B2B SaaS Talks. Lucas and Luna unpack the new enterprise procurement requirement that every AI vendor must provide a model card — a standardized one-page document detailing model architecture, training data, performance benchmarks, and known limitations — before a deal can close. They examine the real-world case of a mid-market AI startup that lost a seven-figure contract to a competitor because its model card disclosed a 3 percent accuracy drop on underrepresented dialects. The episode covers the origins of model cards from Google's 2019 paper, how procurement teams now use them to compare vendors side by side, and why failing to maintain an up-to-date model card can halt a deal in the final approval stage. Lucas and Luna also discuss the tension between startup agility and enterprise compliance, and offer practical tips for building a model card that accelerates sales cycles rather than slowing them down. This is a must-listen for any B2B SaaS founder, sales leader, or product manager navigating the new AI procurement landscape.