---
title: "Auto-essay closes the loop each week"
created: 2025-10-15
visibility: public
status: live
tags: [writing, automation, synthesis, weekly-review]
aliases: ["weekly essay", "auto synthesis", "weekly wrap"]
updated: 2025-10-15
summary: "Longform should be a by-product of work, not a second job. The agent scans the ledger, clusters the week by topic and momentum, and drafts 700–1,200 words in my voice."
---

Longform should be a by-product of work, not a second job. An agent scans [[See what I'm working on]], clusters the week by topic and momentum, and drafts 700–1,200 words in my voice. I can approve, tighten, or skip it.

The essay cites notes inline. Helps readers who won't pane-surf through the full wiki. Helps me see if the week had a real through-line. If not, I know to plan a sharper arc for next week. Essays also surface questions that belong in [[Ask the Brain]].

They expose dead weight. If a section repeats itself, I make a new note or merge two existing ones. The essay turns fragments into a story without breaking the cadence of daily notes. [[Notes beat posts for showing progress]] because the atomic notes stay primary.

The weekly synthesis builds on [[AI shapes voice dumps into atomic notes]]. Raw material comes from daily voice captures. [[I hope people will pay for a working mind]] when they can choose between granular updates and broader narrative context. This automates synthesis work that would otherwise require dedicated writing sessions.

Improves [[Titles are takeaways, not labels]] too. Unclear titles become obvious when the agent tries to cluster related notes for the weekly narrative.