Your job now is to use GPT 5.6 Sol medium in anger.

Fireside infra is all defined in markdown files. Codex can spin up a new app server by just reading the file and cooking. Works great. Unsure how that would work for framework but seems possible.

After three days of cyber cabs in Austin, driving feels gross. 😆

From @tenderlove’s keynote. I’m still here for reading code as well. Not all code but key/important code. Hot paths. Critical areas.

I’ve been doing ruby for two decades (👴)! Finally got to say thanks in person to matz last night. 😊

If you put rust on the rails the train can derail. That is too dangerous.

“I’ll start every new web application with Ruby on Rails still” @dhh in chat with matz Everyone needs to chill out about rails being dead.

AI World 2027. Just cover up the R, L and S.

It’s true. @yukihiro_matz has been vine coding for 15 years. 😂

I did not have dhh quoting the bible on my bingo card. 😂

Normally we get a keynote at Rails World that gives us a new Rails release and some new features. This year we didn't. @dhh spent a good amount of time talking about pencils down, manually written code is dead etc - I think a lot of that is true! It at least lines up pretty well with my experience over the last 6 months. He also spent a greater amount of time talking about the rewrite of Hey in Rust, instead of Rails. But that doesn't mean that Rails is dead, in fact he mentioned that there are still plenty of use-cases out there for Rails, Basecamp being one of them. He emphasized the focus on AI evals, to figure out where you get the most bang for the buck when it comes to Rails, and how great the web as a platform still is. It's easy to be somewhat pessimistic/afraid of the future and see his Rust rewrite story of Hey as Rails being dead - but instead I think it's worth being optimistic, and think back to when all code was written by hand - where, for a Rails shop it wasn't feasible to write native apps. That's how we ended up with web apps, running in web views and a somewhat subpar experience compared to the native implementation. But it was close enough, and gave us a decent enough experience. But in the day where agents can write your code, I think it'd be idiotic not to pursue a native app. Be ambitious! Yeah you might not understand a single line of the code it writes, and maybe that's important to you - then don't do it. But if it works for someone else, I don't think we need to judge them. There's already enough judging between Rails, JavaScript and what not. Don't yuck someone else's yum! And I think the most important point is the positive look at things vs the negative one. We can't always plan for the future, and a negative outlook will impact your day to day - focus on the positives instead! But I also understand the worry as the world around us changes, as less and less freelance clients need your help, as your full-time job becomes more uncertain due to factors you don't control. It's hard! But we shouldn't spiral if we can avoid it. And last, we could've been given a regular keynote, that pretended like nothing had changed in the 12 months since the last Rails World - but that wouldn't be fair either. David doesn't change history or how things plays out, so whether he gives us the truth or sticks to the regular keynote - doesn't change the outcome. But I'd much rather hear the tougher truth, than the "everything is the way it used to be". Use AI to empower you, go out and create the silly project you've already dreamt of, and maybe even dip your toe in the waters outside of your comfort zone and try and build something that isn't Ruby, or Rails, or the web - it might turn out that it's fun? :) #railsworld

Then and now #railsworld

It is over. Writing code is dead. Now we’re makers/builders. Agreed.

Ruby was 55%. Now it’s 3%. Rust on rails here we come.

I’ve been pencils down since December. It’s great.

Rails is dead.

92% reduction in feed serving. 📉 Again, the majority of this was just order of operations and actually prompting codex to take a peek under the hood. You can probably do stuff like this to about any endpoint that hasn't been meticulously optimized. Even on flipper cloud telemetry processing I just cut many millis just be switching from create! to insert_all to avoid begin/commit for a single insert. Duplicate prevention was retained too.

Thanks to my daily Codex/Flare app perf report I'm fixing some stuff on Fireside. Saw slowness. Kicked off a new task. Codex whipped together a 5 phase attack. Been deploying one at a time. Biggest improvement came from a couple line change of when we return 304 not modified. Avoids several queries.

Starting to really take advantage of scheduled tasks in Codex. My App Performance one is really paying off. So I added more things to it (from railway and aws). Yesterday I created an OSS maintainer task to help me not let issues/pull requests slip by. Thinking Stripe is next.