Capture notes automatically from email, meetings and transcripts

Every good PKM habit dies at the same step: moving material from where it arrived into where it belongs. This is the workflow that removes that step.

Capture is where most note-taking systems quietly fail, not because people stop having things worth writing down, but because the path from "this happened" to "this is filed" runs through you, by hand, every single time. An email needs to be read, understood, and re-typed as a note. A meeting needs to be summarized. A transcript needs to be read for the parts that matter. Do that enough times and the system stops being worth maintaining, and the tool gets blamed for a failure that was actually in the workflow around it.

This is build-list item two of the book: capture without copy-paste. One workflow, three inputs: email, meetings, and transcripts, all landing the same way.

It is worth naming why copy-paste is the actual enemy here, not laziness. Every manual transcription step is also a chance for the material to get filed under the wrong project, summarized inconsistently, or simply never get around to at all on a busy day. Removing the step does not just save time. It removes the point where the whole habit is most likely to break.

What triggers it

Material arrives in any of its raw forms: an email thread, a meeting transcript, a call recording, a document, or an article worth keeping, covered separately in save articles for AI.

What the system does

1. Reads the material in its native form

No manual retyping, reformatting, or summarizing by hand first. The system reads the email, the transcript, or the document directly.

2. Identifies what it actually is

A decision, a status update, a piece of reference material, an action item, a receipt. What kind of thing it is determines where it goes next, the same classification used across email triage and meeting notes.

3. Matches it to the right project or person

Checked against your existing project and contact files, the same way you would recognize where it belongs from context, rather than needing you to specify it up front.

4. Files a dated note and updates whatever it changed

The raw material, or a summary of it, gets saved dated inside the right place, and if it changed a project's status or added a decision, that gets updated too, not left for a later pass.

What lands in front of you

A dated, filed note for whatever came in, already attached to the right project or person, plus any downstream update it triggered, a new decision logged, a status changed, a follow-up drafted.

What you still decide

Whether the system filed something under the right project, and, for material it was not confident about, where it actually belongs. Everything else, the reading and the filing, is off your plate. Over time the number of things you have to correct drops, not because the system got smarter on its own, but because your projects and contacts are better defined the longer the hub runs.

The capture layer under everything else

This is the shared mechanism behind three more specific workflows. Obsidian AI plugins can bolt a chat window onto a vault, but do not do this kind of routing on their own, and Apple Notes vs Obsidian covers the plain-note side of the same question.

This capture layer is what feeds meeting notes and transcript-to-action-items as more specific downstream cases.

Treat those three as specializations rather than separate systems. Underneath each one is the same read, classify, match, file sequence described above, tuned for the particular shape of the input. That is deliberate: learning one capture workflow means you already understand all of them, which is a large part of why the system stays maintainable as more of it gets built out.