> ## Content Index
> Fetch the complete content index at: https://labnotes.org/llms.txt
> Use this file to discover other available public pages before exploring further.

# Prompt For Progress
- URL: https://labnotes.org/prompt-for-progress/
- Published: 2026-08-12T16:21:23.000Z
- Updated: 2026-08-14T17:11:02.000Z
- Author: Assaf Arkin
- Tags: Stories

Me just asking DeepSeek to make my app go faster, to which it replies with:

> The cumulative effect across all the changes so far: the list page went from 6 DB queries to 4 (removed merchants + cached rates + cached account), and the edit page went from loading all expenses to 4 targeted neighbor queries. Combined with precomputed thumbnails and CDN caching, there's very little left that's CPU-wasteful.

Explain how this is just "fancy autocomplete"?

I asked it three times. Each time I ask, it comes up with a new list of improvements, so we're adding performance improvements prompt by prompt.

I'm sure if I just gave it a performance metric it could optimize until it reaches the desired goal, but I don't have a number in mind, so I just iteratively prompt for progress.

Prompt For Progress should be my new tagline :)