We want fast verifiers, but we also want to write as few specifications as possible. Liquid type inference has long been used by auto-active verifiers such as \textbf{Flux} or \textbf{LiquidHaskell} to reduce annotation overhead. Even though this inference \emph{can} be used to infer some function contracts, most verifiers like \textbf{Flux} instead require users to spell out the contracts themselves to enable modular — and thus scalable — verification. We discuss our efforts to reduce user annotation overhead \emph{without} compromising on modularity or speed.