As I'm sitting here, on question 31 of a /grill-me prompt, having asked GPT 5.5 to "make db path configurable", I can't help but marvel at the endless void of pedantic decisions we devs have been expected to make daily. Hard to imagine a non-dev surviving a /grill-me session.

I scan and shred lots of docs (kids' schoolwork, mail, etc), so wanted a simple way to let AI at them. Vibed this up and been using it happily for a couple months now. Maybe useful to someone else. Contrib welcome! github.com/maxim/ringbi...

Tried to have AI read my posts and codify the style, then made several different LLMs try to write as me. The result is unfortunately beyond cringe. Weirdly, simpler models (e.g. Qwen 3.6-plus) came the closest, but it's still painful to read. Hard nope.

A new way to recognize AI-built web frontends: it's broken in Safari. As a Safari user, this is getting old fast. I thought AI should help cover more ground for you?

Using plannotator with pi lately (liking it so far). Wrote a seeds file for my Rails app. Code review from Claude: "here're a couple of nits". Code review from Codex: "you're about to break production, because you're doing db:prepare which also runs your seeds".

Something's missing in these gstack-style approaches to AI. Why reupload millions of tokens of its own slop back into it? It's in there already. All it needs is what it doesn't have: your data, and some steering to apply its knowledge according to your preferences. Use pi too.

Added pi to the list. Unironically considering it as the base for AI setup. bsky.app/profile/max....

PSA for those caught in the AI hamster wheel, something I wrote in my college notebook 22 years ago, still holds up: Happiness is a cause, not an effect. Get happy now, in your mind, then do things. You don't work towards it, you start with it, and it enables everything else.

Here's where the puck is going: AI will replace software. You ask it to solve problems, you instantly get custom apps. All computing will become this. We will be sharing T=0 prompts that produce useful results. All software engineering will become about improving AI itself.

My attempt at deciding on an AI orchestrator. So many out there. Suggestions/corrections welcome. gist.github.com/maxim/c6ef7f...

If you're gonna embrace non-determinism at all stages of software dev, including verification, why even make software? Just have AI be the software. Why have an "if" condition when AI can decide. Unironically, that's where we're going. It's moving ground all the way down.

Of all the AI artifact-managing solutions I've seen so far, beads is the one that nails it.

Is it just me, or @github.com's PR reviews have been getting more and more buggy? Now when editing a pending comment, all formatting is lost. Bullet points, paragraph breaks, `code` quotes. You just get a chunk of plain text and have to redo the formatting by hand.

If you've seen my blog post 4 years ago titled "Don’t Build a General Purpose API To Power Your Own Front End", I've finally finished writing a "4 years later" follow up to tell how it played out, and answer some piled up questions. max.engineer/server-infor...

Everything seems to have broken. @topazlabs.bsky.social is failing all my family video renders (for days now). commento.io is down (expired domain?), so can't keep working on my response-to-comments blog post. @simpleanalytics.bsky.social won't provision SSL for my domain. Wtf is going on internet?

MacOS >=13 hides accurate AI-based OCR called LiveText. It's very good, it's what lets you select text in images. I was working on a Ruby-Swift bridge for it, to make scanned PDFs searchable, but a few hours ago this project beat me to it. Much recommend! github.com/mkyt/OCRmyPD...

Most software has the same relatively simple data transforms, under different names and reasons. Could it be that if we (developers) focused less on business jargon in types and hierarchy, and more on the story of data transformation, we would get code that's easier to follow?

Which AI does stock market best. A fun thing to observe. hkuds.github.io/AI-Trader/ (GitHub here: github.com/HKUDS/AI-Tra...)

When you ask LLM for sources, and it gives them to you, you better damn well read them. This just happened yesterday in a codex session: news.ycombinator.com/item?id=4575...

A controversial Monday thought for you: ~15 years of experience is when a programmer is the most dangerous for the ecosystem. That's when you beautifully wrap up your subpar experiments with a pretty bow on top for others to use, and they actually gain traction.

A thing that bothers me about total AI takeover narrative is the perpetual need to have its back "just in case", and what that actually implies. max.engineer/failover-to-hi

After considering elaborate ways to make TinyMCP support different return formats, finally arrived at this "genius" 3 line solution, which brings you version 0.2.0. Keep it simple folks. github.com/maxim/tiny_m...

UX issue on Github: if you search, there is no way back to where you were before. You must either back-back-back in the browser, or enter the original url again. It's interesting, because it feels super annoying in practice, but actually not obvious from webdev standpoint.

Even 1-shot-correct code by Claude 4 is subtly bad. Gotta read it with fine tooth comb when project matters. What's worse, "instructing LLM" brain mode feels mutually exclusive with "writing code" brain mode. It hurts to switch (and I'm an easy context switcher).

Published my local tiny_mcp server that I use to create quick simple MCP tools in Ruby. github.com/maxim/tiny_mcp

Impressive how easy it is to write an assistant that reads and answers questions about a >1200 page PDF with RubyLLM. Less than 20 mins of work. max.engineer/giant-pdf-llm

Is it just me, or reading text in a different style/format than it's written causes the "fresh eyes" effect? That's why I like to review code and blog posts in a different viewer than my editor. It's as if I'm reading it a day later, without actually waiting a day.

Every "services" gem: do you dislike this code? [Shows a plain ruby procedure in a controller action]. Here's how you could write it instead: [Shows a lambda-heavy mix of actors, executors, policies, contexts, and almost the same procedure in the controller action].

Remembering a childhood story in the shower: So it's D'Artagnan and the Three Musketeers: Athos, Protoss, and Zerg.

ActiveRecord supports optimistic locking, but it requires the `lock_version` column. It'd be nice to save a record with a WHERE constraint and callbacks, but w/o needing that column. Here's a proposal: https://discuss.rubyonrails.org/t/proposal-optimistic-locking-without-lock-version/83221

This little trick has been keeping rubygem monkeypatches from breaking across a number of my jobs: https://notes.max.engineer/gempatch-monkeypatch-rubygems-responsibly

For some reason I couldn't find any quick/easy examples to create a table and a model inside a Rails test. So I threw one up onto my notes site in case you need it. https://notes.max.engineer/helper-for-creating-table-and-model-in-rails-tests