Towards GeneraliZable Implici | Pangram Labs