In a short Office Hours conversation, Bardia Shahali contrasts sales in AI infrastructure versus AI applications. Shahali built sales at Fireworks AI from roughly $3m to $400m in revenue and is now VP of Sales at Granola. The core takeaway is that infrastructure deals are large and fast — an eight-figure deal can close in two weeks because customers are AI-native companies growing at extraordinary rates. The trade-off is fierce competition: every inference provider chases the same accounts, so stakes are higher, but each rep’s quota is multiples higher than in other markets.

Shahali also says the pace of change breaks traditional planning. Instead of annual comp plans, his team now sets comp every quarter because products evolve quickly, model capabilities shift roughly every 45 days, and the competitive landscape moves fast. Understaffing is the third risk: with a small team, you don’t put in extra effort for outbound, which hurts expansion.

At the infrastructure layer, forward-deployed engineers (FDEs) and sales engineers (SEs) pair with account executives (AEs) to deliver a highly complex product. At the application layer, AEs and customer success managers (CSMs) deliver software that resembles the previous era but with higher productivity and faster adoption/expansion. The full 15-minute conversation also covers on-target earnings in the AI era, onboarding, burnout, and usage-based pricing at the app layer.