By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
At a glance
WHAT IT’S REALLY ABOUT
Figma’s internal PM OS: skill repo, bug bashing, AutoXP experiments
- Ezra Mechaber demonstrates Figma’s internal “PM OS,” a shared repo of ~31 AI-enabled skills designed to help PMs automate repetitive work and raise baseline execution quality.
- He explains PM OS’s structure: an agents.md for behavior, skill files and hooks, plus a shared parent repo and private sub-repo workspace so many PMs can stay in sync while keeping personal work private.
- He highlights everyday workflows like auto-identifying relevant collaborators and a “what’s up” Slack synthesis that aggregates channels, decisions, and obligations to help PMs focus amid information overload.
- He shows how usage shifted from a terminal-based Claude setup to a local web UI and then largely to Codex due to improved browser automation, including using Codex for systematic bug bashing and reporting results back into FigJam.
- He introduces AutoXP, an automated experimentation system that coordinates data science, UX writing, and engineering agents via Slack-triggered automations to launch and analyze more experiments faster, and shares a prototype real-time writing assistant that uses Jev to flag missing support/claims and pull substantiating quotes from a library when the writer pauses.
IDEAS WORTH REMEMBERING
5 ideasPM OS started as a push to operationalize AI productivity for PMs, not just engineers.
Ezra Mechaber said he built it after realizing few non-engineers at Figma were using AI tools to get real work done, and after a challenge from then-CPO Yuki about what it would take to get PMs “AI pilled.” The intent was to turn a personal, terminal-based workflow into something reusable for many PMs.
A core goal is “raise the floor” on neglected PM tasks with minimum-viable quality.
Ezra described the driving problems as PMs doing too many jobs and missing or underperforming on key tasks (like looping in support at launch time or producing usable meeting notes). He framed the goal as getting to “minimum allowable” quality so PMs aren’t blocking others.
PM OS is a lightweight, Git-based skill library with shared + private layers.
Ezra explained the system is a basic repo with an agents.md (behavior), a set of skills (when to fire/what to do), and command hooks (like checking for updates). It’s structured as a shared parent repo plus a private sub-repo workspace so individuals can write back privately while keeping a common template in sync.
The most-used skills center on org context + Slack triage to manage attention.
Ezra highlighted skills that reduce repetitive context work—like auto-including the right collaborators in docs (e.g., EM, designer, relevant reviewers) and identifying who owns what in the org. He also uses a Slack-focused “what’s up” skill multiple times daily to synthesize channels, projects, outstanding decisions, and obligations.
Adoption and iteration were driven by feedback loops, not instrumentation.
Ezra said the team intentionally avoided telemetry because he didn’t want people to feel “creeped,” wanted it to stay fun, and didn’t have time since it was “20% time.” Instead, they relied on heavy feedback loops via chronically-online PMs, a public channel, and adoption by other roles (engineering managers; a data science fork).
WORDS WORTH SAVING
5 quotesI like to tell my engineers that I... my goal is to be lazy in the right places.
— Ezra Mechaber
I, as a growth PM, I feel like the biggest sin you can commit is wasting traffic, right? To, like, have unallocated traffic that is not running towards an experiment at any given time is like, why are you wasting that traffic?
— Ezra Mechaber
What if we could just automate myself out of a job and, like, create a little growth PM who could just, like, run and run and run.
— Ezra Mechaber
I archive roughly 10 channels a day. It's bananas.
— Ezra Mechaber
I don't think they're conscious. I don't think they're approaching conscious, but I do think it's really bad for the soul to just be mean to things all the time.
— Ezra Mechaber
High quality AI-generated summary created from speaker-labeled transcript.
