Main idea

Three families of material feed a product workspace, and they do not call for the same gesture.

Documentary curation — books, articles, standards, product documentation, reference frameworks — calls for a human choice, but it bears on objects that already exist and remain available: the job is to spot them, add them, keep them. Industrialised user research — interview transcripts, in-app feedback, support tickets, verbatims from a CRM — lends itself to automation, because those traces are produced by a machinery that runs without anyone thinking about it.

The third family has no producer at all. What the product manager knows, understands, senses or connects exists nowhere as an object: no tool emits it, no export contains it, no indexing will find it. There is therefore nothing to collect — what is needed is a channel through which someone deliberately pours it in.

Confusing these three families produces bad setups: you expect a connector to go and fetch material nobody ever emitted.

Why it matters

This says where to put the investment. On the first two families, the effort goes into tooling and shrinks over time; on the third, the effort is a repeated human gesture, which never shrinks and which no integration buys back.

It also gives the reason a workspace plugged into every tool in the company can stay poor: it has industrialised what was already industrial.

Nuances and limits

The boundary moves. Systematically recording customer calls moves part of what belonged to the third family into the second — without moving the interpretation, which still has to be poured in.

And "unwritten" does not mean inaccessible: part of this material comes out naturally in conversation, provided a place exists to receive it at the moment it comes out.

Open questions

• What share of a product manager's unwritten knowledge stays out of reach even with a dedicated channel, because it cannot be put into words?