AI Amplifies What's Already There
What I took away from Tropical Rails 2025.
I went to Tropical Rails with the usual expectations: see what the Ruby community is up to, check out new tools, talk to people. I didn’t expect the entire conference to orbit around a single topic. But it did. AI and agents. Not in the “check out this new tool” sense, but in the “this changes the rules and we need to talk seriously” sense.
The takeaway
The conclusion I walked away with, and the thread running through almost every talk, is simple but forceful.
LLMs and AI agents don’t have opinions about code quality. They amplify what’s already there. Good patterns in, good code out. Tech debt in, more tech debt out. Faster.
If the codebase isn’t clean and prepared to scale, it won’t have the capacity to keep up with product growth. And now that agents can produce code at a pace no human can review, that’s not a quality problem. It’s a capacity problem.
It’s not that this wasn’t true before. It was. But it matters more than ever now.
Two tools worth pinning
Two gems came up multiple times and are worth mentioning.
evil-seeds, like a database dump but smart. You can configure how data gets scrubbed, which tables to include, how many records to pull. Sounds like a good path to finally having a clean, safe copy of production to work with locally.
Blazer, for spotting weird or inconsistent data in production. The idea of having something that actively monitors data health is something we could use, whether or not we adopt this specific tool.
The real takeaway
If we’re serious about using AI to move faster, the codebase needs to be in a state that makes that possible. Tech debt isn’t just a quality issue anymore. It’s a capacity issue. And agents amplify it. They don’t fix it.