---
title: "Agents change the math on big bang migrations"
url: https://daily.dev/posts/agents-change-the-math-on-big-bang-migrations-pac5lf3qj
source_url: https://swizec.com/blog/agents-change-the-math-on-big-bang-migrations
type: article
source: "swizec.com"
published: 2026-08-01T16:51:38.604Z
updated: 2026-08-01T18:18:16.802Z
tags: ["webdev", "tailwind-css", "ai-coding", "shadcn"]
reading_time: 7
upvotes: 34
comments: 8
language: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Agents change the math on big bang migrations

**[swizec.com](https://daily.dev/sources/swizec)** · 7 min read · 34 upvotes · 8 comments

## Summary

A senior engineer shares how AI agents (Claude) made a previously impractical big-bang migration feasible: migrating hundreds of pages from Bootstrap to Tailwind CSS in 5 days with a 22,000-line PR. The post walks through three experiments — an unguided yolo rewrite that produced a mess, a guided rewrite with clear design system direction that mostly worked, and a cleanup pass that reduced class token count from 33K to 12K. Key insight: AI can handle mechanistic transformations at scale, but humans must define what 'good' looks like first. The author estimates the whole effort cost a few hundred dollars in tokens and would have been economically unjustifiable without AI assistance.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://swizec.com/blog/agents-change-the-math-on-big-bang-migrations>

## Community discussion

Top comments from developers on daily.dev.

**@rizzdev** · 2 upvotes

> Who debugs the next incident in 22k lines nobody has read?

**@pdfopsdev** · 2 upvotes

> 33k to 12k class tokens after the cleanup pass is the real story here, not the 5 day timeline. that's the part people skip when they show off the yolo-rewrite diff stat, the guided pass plus manual cleanup is where the actual maintainability wins live and it's still mostly human judgment calling the shots.

**@trevorsuna** · 1 upvotes

> The cleanup pass is the strongest evidence here: agents made the broad rewrite affordable, but the design system still had to define what good looked like. I’d pair a migration this large with visual regression tests and incremental deploy flags even if the code change lands as one PR.

**@kartiknvj** · 1 upvotes

> The 22k-line migration in five days is impressive, but the line that mattered to me is that humans had to define "good" before the agent ran. That's the whole game: the agent can execute at scale, but if the target state is fuzzy you just get 22k lines of confidently wrong. How did you encode "good" concretely enough that the agent could check itself against it?

**@vikyw** · 0 upvotes

> Casino math ?
> Martingale ?
> Parallel bet ?

## Similar posts on daily.dev

- [How I Used AI Agents to Migrate my Entire Site to Astro](https://daily.dev/posts/how-i-used-ai-agents-to-migrate-my-entire-site-to-astro-ba6sfp5qn) · Code with Andrea · 1 upvotes · 0 comments

---

Tags: [#webdev](https://daily.dev/tags/webdev), [#tailwind-css](https://daily.dev/tags/tailwind-css), [#ai-coding](https://daily.dev/tags/ai-coding), [#shadcn](https://daily.dev/tags/shadcn)

[View this post on daily.dev](https://daily.dev/posts/agents-change-the-math-on-big-bang-migrations-pac5lf3qj)