Skip to content

Commit fe50542

Browse files
committed
feat: add new link 'Eight years of wanting, three months of building with AI'
1 parent 800d313 commit fe50542

1 file changed

Lines changed: 40 additions & 0 deletions

File tree

  • www/links/2026/04/07/eight-years-wanting-building-ai
Lines changed: 40 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,40 @@
1+
---
2+
tags:
3+
- article
4+
- link
5+
author: eric
6+
linkTitle: Eight years of wanting, three months of building with AI
7+
date: 2026-04-07 21:10:01 -07:00
8+
linkUrl: https://lalitm.com/post/building-syntaqlite-ai/
9+
viaUrl: https://simonwillison.net/2026/Apr/5/building-with-ai/
10+
teaser: >-
11+
Lalit Maganti and I have a similar experience with the need to constantly refactor when
12+
building a project with a coding agent.
13+
---
14+
15+
Lalit Maganti wrote a [remarkable account][link] of building [syntaqlite] using Claude Code:
16+
17+
> If you’re using AI to generate code at industrial scale, you _have_ to refactor constantly
18+
> and continuously. If you don’t, things immediately get out of hand. This was the central
19+
> lesson of the vibe-coding month: I didn’t refactor enough, the codebase became something I
20+
> couldn’t reason about, and I had to throw it all away. In the rewrite, refactoring became
21+
> the core of my workflow. After every large batch of generated code, I’d step back and ask
22+
> “is this ugly?” Sometimes AI could clean it up. Other times there was a large-scale
23+
> abstraction that AI couldn’t see but I could; I’d give it the direction and let it
24+
> execute.
25+
26+
The whole post is worth reading. With the project I have been working on I have been having
27+
similar experiences. Especially the need to always prioritize refactoring.
28+
29+
My current project recently reached the point where it was getting hard to reason about
30+
areas of responsibility, so I decided it needed to be refactored into a monorepo. In non-AI
31+
time, this is something I would have sat on for a while longer before begrudgingly taking it
32+
on. But in AI-time, delaying this refactor would mean compounded deterioration of my
33+
understanding of how the pieces of the codebase are interconnected.
34+
35+
Currently, I’m keeping a `BACKLOG.md` file in my project. It’s just a handful of sections
36+
with bullet lists of to-dos. Right at the top is a section for refactors. If I think of a
37+
refactor it becomes the very next thing to work on. I have yet to regret doing a refactor
38+
before doing more feature work.
39+
40+
[syntaqlite]: https://lalitm.com/post/syntaqlite/

0 commit comments

Comments
 (0)