Lenny's Podcast: a Whatnot CPO on why product management shouldn't exist — and what it should become

Tom Verrilli is the Chief Product Officer at Whatnot, the live commerce platform, and has held product leadership roles at Twitch and Twitter. He appeared on Lenny Rachitsky’s podcast in August 2026 for a conversation that starts from a pointed premise — that Verrilli regrets product management as a discipline exists — and uses that frame to examine why traditional PM structures underperform and what AI changes about that equation.

Who it’s for: product managers at any level who are questioning whether the structure of their work reflects how products actually get built today, and product leaders evaluating how to reshape their teams as AI tools absorb execution overhead. Also useful for PMs at organizations with large coordination-heavy PM functions, where the pressure to restructure will be most visible.

Key takeaways:

1. The case for fewer, more senior PMs is not primarily about headcount reduction. Verrilli argues that PM value comes from hands-on product judgment — building, prototyping, making things testable — rather than from owning process or managing information flows between functions. His team at Whatnot is deliberately small and senior, and expected to build things directly. The argument is that large PM functions built around coordination tend to accumulate process overhead that slows product decisions rather than improving them.

2. AI changes the unit of PM work, not just its speed. Verrilli describes how AI tools have compressed the distance between a product hypothesis and something testable. Where a PM team previously spent a week moving from insight to validated experiment, the same loop now takes a day. This changes how many hypotheses a small team can pursue in a quarter, which changes how you think about team size and specialization. It is not simply “write the PRD faster” — it is a different cadence for the entire discovery-to-experiment cycle.

3. The “play the accordion” model describes a skill that becomes more important, not less, as AI handles middle-layer tasks. Verrilli uses this metaphor for the ability to move between very detailed execution work and very high-level strategic framing quickly and repeatedly. As AI absorbs more of the synthesis and documentation work between those two levels, PMs who can only operate in the middle tier become less necessary while those who can move across the full range become more valuable.

4. Organizational design should follow what your team can actually do with AI, not inherited ratio norms. Verrilli is direct about this: how Whatnot structures product work is a product of specific decisions about tools and workflows, not a result of applying a PM-to-engineer ratio standard. His implicit challenge to leaders is to design around demonstrated capability rather than conventional headcount models.

5. Data science and product work are operationally converging at Whatnot. Verrilli discusses how AI has changed the data science function in tandem with the PM function — the work of generating insights, running experiments, and interpreting results is changing at a similar rate to product work, and the interaction between the two is where a significant amount of new capability is being built.

Worth watching if you have been adding AI tools to existing workflows without changing how the team is structured or what the PM role is actually supposed to produce. Also relevant if you lead a product organization and want a concrete framework for thinking about where AI-era PM value actually comes from.