EPISODE 30: Microsoft product leader Sangya Singh joins Jason and Kevin to unpack how to decide what to ship first in AI and automation. She shares a “strategy to win” playbook (fall in love with the problem, define the hypothesis, then hire and build), why agility must be daily not monthly, and how Microsoft balances agentic and deterministic systems—highlighting a risky-but-breakthrough bet on self-healing RPA. The crew contrasts outputs vs. outcomes, explores eval-driven prioritization, and talks scale mechanics inside Microsoft. Sangya closes with what’s next: voice-based AI surfaces that discover what to automate and “mech-interrupt” style safety tooling so enterprises can see, govern, and correct model behavior.
