About me

Peter Farnham
Product lead · Apple · Meta
Fifteen years of product at Apple and Meta. Three keynote launches, features in two billion pockets, a decade in the rooms where the roadmap actually gets decided.
- 15
- years in product
- 3
- keynote launches
- 2B
- pockets reached
Where I came from
I started the way most product people start, which is by being wrong in public. Apple taught me the opposite habit: you do not ship the thing you argued for, you ship the thing that survived the room. Three of those made it all the way to a keynote stage. The one I am proudest of is the one nobody noticed, because it was supposed to be invisible.
Meta taught me scale, which is a different discipline entirely. At two billion people every decision has a tail. The edge case you wave off in a Tuesday review is a city's worth of humans by Friday, and you stop designing for the user and start designing for the distribution.

What I think is coming
For thirty years the bottleneck in software was construction. Somebody had an idea, and then a team spent two quarters turning it into a thing. That bottleneck is gone. What took a team a quarter now takes one person an afternoon, and the constraint has moved somewhere far more interesting: taste. When anyone can build anything, the last scarce input is knowing what is worth building.
The interface moves next. I spent a decade fighting over pixels, and screens are about to stop being where the fight is. The coming interface is intent. You will not learn a product, you will describe an outcome and watch the software assemble itself around it. Most of what a product manager does today is translation between what a person wants and what a system can do, and that translation layer is exactly what is being automated. The people who stay useful are the ones who know what to want.
Then distribution inverts. You used to need an app, a store, a download and a retention curve. The smallest shippable unit now is a paragraph somebody pastes into a chat they already pay for. No install, no account, no permission from a platform. That is the biggest thing to happen to software distribution since the URL, and almost nobody is building like it has happened yet.

Why this library exists
If the scarce thing is knowing what is worth making, then the useful artefact is not another model or another tool. It is a collection of judgements that already survived contact with reality. That is what this is: the notebook I kept for myself, opened up.
So nothing enters on a press release. Every prompt, file and reference here was run first, and it stays only if it works on the first paste, in whatever AI chat you already pay for. That is the whole editorial policy, and it is harder to hold than it sounds.