Write As You Go. Schedule Ahead.

Source context: This field note was first shared through Eric's Advisory Hour Substack notes stream. View the original note.

Infographic comparing the old way of rushing to write after a project with the better way of capturing lessons and scheduling posts ahead — Write as you go, schedule ahead, and turn the work itself into the capture process.

The note

Confession: I used to write on the same day I finished a project. Now I do the project and write ahead, capturing lessons as I go.

Scheduling ahead is freedom and a learning loop. It sounds simple-but it isn’t obvious, or wasn’t to me.

Scheduling ahead as a concept? Not once did I have the idea. I just never once thought to schedule posts. Lord, how sometimes the obvious step isn’t so clear until you’ve traveled the harder path.

Sharing because if you’re the kind of person that writes when the project is “good enough to be done”-don’t. Don’t wait-it’s what creates the anxiety and stress! Try flipping the script and write as you go.

Then on busy days, your readers are served. And even better? You have something of value: insights not yet broadly shared you can send to a select few. It’s real-you’ll find yourself sharing “from the future” to a select few.

If you don’t understand this process, and why it works. I had this image made. Good luck fellow garage tinkering project writer!

Why this belongs in the SLM field notes

The same loop that makes publishing less frantic also makes small-model work more useful. The best artifacts are usually captured while the experiment is still running: what failed, what surprised, which prompt shape worked, which eval caught a regression, and which tiny model behavior is worth training into a repeatable ability.

For the Custom Model Training / SLM thread, this is the operating habit behind the technical notes: build the thing, capture the lessons as they happen, and schedule the write-up from the fresh evidence instead of trying to reconstruct it after the context has gone cold.

See the surrounding Custom Model Training notes for the hardware, nanochat, pico-model, and ability-training side of the same practice.

Original source

This local copy preserves the note text, source link, and inline media. Canonical Substack note URL: https://substack.com/@advisoryhour/note/c-266148171.