Coding with LLMs in the summer of 2025 – an update

Sorry if I missed it in the article — what’s your setup? Do you use a CLI tool like aider or are you using an IDE like cursor?

He uses vim and copy paste code from web interfaces because he wants to maintain control and understanding of the code. You can find proofs of this setup on his youtube channel [https://www.youtube.com/@antirez]

Terminal with vim in one side, the official web interface of the model in the other side. The pbcopy utility to pass stuff in the clipboard. I believe models should be used in their native interface as when there are other layers sometimes the model served is not exactly the same, other times it misbehaves because of RAG and in general no exact control of the context window.

Thanks. Also based on the coding rig you use models may not match the performance of what it is served via web. Or may not be as cheap. For instance the Gemini 2.5 pro 20$ account is very hard to saturate with queries.

This seems like a lot of work depending upon the use case. e.g. the other day I had a bunch of JSON files with contact info. I needed to update them with more recent contact info on an internal Confluence page. I exported the Confluence page to a PDF, then dropped it into the same directory as the JSON files. I told Claude Code to read the PDF and use it to update the JSON files.

It tried a few ways to read the PDF before coming up with installing PyPDF2, using that to parse the PDF, then updated all the JSON files. It took about 5 minutes to do this, but it ended up 100% correct, updating 7 different fields across two dozen JSON files.

(The reason for the PDF export was to get past the Confluence page being behind Okta authentication. In retrospect, I probably should've saved the HTML and/or let Claude Code figure out how to grab the page itself.)

How would I have done that with Gemini using just the web interface?

Do I understand correctly that you deliberately entered personal contact information into LLM?

If so, I would be a reprimanding anyone in my org that did this. While it’s more effort I’d use the LLM to write a script to read the page with the Confluence api, parse it, write out the json files and push them where they need to go.

Add in basic assertions to check the data is present, in the expected format and there is enough of it. Alerting when the assertions fail, then I can schedule it and forget about it.

This is where LLMs shine, I can now build a robust solution in an hour instead of a day.

I’ve seen this exact workflow (PDF → extract data → update structured files) come up a lot, and it’s impressive that Claude handled it end-to-end like that. We’ve been building Retab.com to handle those kinds of tasks more reliably, especially when you want structured output (like JSON) from messy documents like PDFs, scans, or even images. Instead of writing ad-hoc scripts or chaining LLM calls, you just upload the file, define what you want (via schema), and it gives you clean structured data. It’s AI-native but deterministic, no need to install PyPDF2 or debug model behavior. Just wanted to share in case others are solving similar problems repeatedly.