中文
Cursor Compile

Engineers ship 8x more code and demand hasn't followed — the PM's new ground is commercial value, novel vision, and bringing people along.

The New PM, Claire Vo | Compile 26 · Claire Vo

26 min
AI ProductAI Coding

26 min total·Actually worth watching closely: ~7 min·2 must-watch clips

Orange = the 7 minutes worth watchingFor the rest, the guide is enough
Segment guide · 7 segments
  1. 0:00 2:10Listen

    Opening: what's left of the PM when anybody can build anything

    Claire Vo sets up the talk's core question — what it means to be a product manager in an age where anybody can build anything — using the phrase loosely: product person, product agent, product autonomous loop, whatever we're calling it. Loose room, jokes about drawing the 11:30 slot.

    Not another 'AI killed the PM' panic — a redefinition of the decision of what to build, why to build it and how to build it

    Spoken setup throughout; the audience moment around 107s is just a laugh line, nothing on screen you need▶ Jump to 0:00
    Speaker · Claire Vo
  2. 2:10 7:08Skim

    Process archaeology: agile and backlogs were inventions for not building

    A tour of the creative ways the industry structured its not-building of things — agile loops, gates you had to pass or fall back below the cut line, backlogs of ideas you weren't sure you had capacity for, PRDs specifying how buttons worked. All of it logical when building was expensive, slow and risky.

    Today's agent loops and autonomous coding are an expression of the exact same thing we've seen before: process in service of building, not of building something people want

    Slides of process diagrams here; a quick scan is enough, the argument carries in the audio▶ Jump to 2:10
    Speaker · Claire Vo
  3. 7:08 8:35Watch

    Capacity is no longer the bottleneck: 8x more code

    An Anthropic chart showing engineers shipping about eight times more code per quarter. What one person can ship is going up significantly, what a team can ship is going up significantly, and it's about to outpace the existing backlog of great ideas.

    The constraint has shifted from whether you can build it to what is worth building

    The capacity chart from 431s is the first slab of data the whole argument rests on — worth pausing to read the axes and the magnitude▶ Jump to 7:08
    Speaker · Claire Vo
  4. 8:35 14:08Watch

    The gap: production is exploding, demand is not

    App Store submissions and app counts have gone up while installs, ratings and use have not. There is just not consumption of our code at scale in the market — output and commercial results have come apart.

    'Production is exploding, but demand is not proportionate to production' is the problem for ambitious builders right now

    At 515s she says 'you all have seen this chart' — this supply-demand gap slide is the talk's central visual argument▶ Jump to 8:35
    Speaker · Claire Vo
  5. 14:08 17:40Listen

    The new job, parts one and two: commercial value and behavior change

    Product management redefined, first two parts: 'what people want' has to decompose into commercial value — will people pay to solve this problem — and into behavior change, the badly underappreciated half, getting 100 and then 1,000 people to log in and actually change how they work.

    Identifying a problem and building a solution aren't scarce; behavior change is, and shipping something of reasonable value almost never produces adoption on its own

    No key visuals in this stretch — dense spoken argument, good to listen closely or even at speed▶ Jump to 14:08
    Speaker · Claire Vo
  6. 17:40 21:14Listen

    The new job, part three: novel vision and bringing people along

    Manifesting novel ideas from first principles is what AI cannot help with — no model confirms a brand-new idea will work. And distribution doesn't end at problem and solution: customers, community, teammates and whole markets have to follow you into your vision.

    When anybody can code, anybody can ship, anybody can build an agent, the moat is who convinces the best talent and convinces an entire market to come along

    Also pure argument with no visual dependency — the value is in the claims, so listening beats watching▶ Jump to 17:40
    Speaker · Claire Vo
  7. 21:14 25:46Listen

    Closing: build a company, not just a codebase

    Titles will survive — organizations and companies are very comfortable with titles — but the job will not. The engineer's job is already different, the designer's job is different, and product folks whose work doesn't feel different from a year ago are heading somewhere bad.

    Product management is not dead, but the job is — and the end state to avoid is a very big codebase and a very little company

    No visual dependency in the close, and the lines land densely — worth hearing all the way through to bind the talk together▶ Jump to 21:14
    Speaker · Claire Vo