How AI PMs Ship Features Users Love (Descript CEO Explains)

Descript mapped creator workflows (scripted vs. improvised) and attached AI to specific pains like retakes, eye contact, and clarity—then hid prompts behind dependable, job-based buttons.

Fixed tools work well when inputs are bounded, but Create Clips requests kept adding parameters; Underlord emerged as the right abstraction once users needed highly customized, conversational control.

Before formal eval stacks were common, the team tested on production-like content and shipped when results were genuinely usable; later they layered regression tests, A/B tweaks, and more automation.

Burkhauser argues only the PM can codify what “good,” “acceptable,” and “harmful” look like because it requires deep job/context understanding (e.g., judging jump-cut density, not just grammar).

Studio Sound quality regressed when evaluators used unrealistically terrible audio; the best model for ‘disaster audio’ differed from the best for the common ‘laptop mic’ use case, so the dataset must match the target workflow.

Underlord’s open-ended input increases weird edge cases, so Descript used private alpha with diverse users (editing/AI novices and experts), built regression suites from real prompts, then expanded to new-user activation and opt-in broader use.

To avoid founder micromanagement spirals, Burkhauser recommends humility, intense product usage, customer closeness, business fluency, and continued excellence in core IC execution while gradually taking on strategy.