In last week’s Gav’s Note I talked about how I revamped my journal routine. Mood and worries do not make the entry anymore, just hard data of the day.
But there’s a gap. How to make these daily notes useful?
Self-review is highly biased. I routinely filter out my own friction points to protect my ego.
I need more than willpower to fix this problem. What if something scans through my daily notes and finds the behavioral and thinking patterns I keep missing?
Here’s how I do it:
Log raw. I write down everything I did for the day, what I think, the decisions I made, my running mileage, and products I built, all in my daily note. Emotional stuff goes under its own diary heading, kept separate.
Write a strip rule. One line in my instructions.md does the real work: only log a pattern if it shows up in at least two separate dated notes, no single day counts. I run it through OpenCode with DeepSeek, but any agentic tool and model works just fine.
Run it on a schedule. Sunday evening, one prompt “Analyze this week”. End of month, one longer pass.
This workflow turns a week or month of scattered notes into one report I never had to write myself.
I revisit Austin Kleon’s notebook this week. The way he flips through old pages looking for what repeats inspired me to build this workflow.
Just another coffee thought from Gav
