Jag minns när alla skrek att Twitter skulle dö när Musk skickade 80-90% av personalen också. Fullt fungerande fortfarande (även om det gick nerför ett tag i början). Definitivt mycket överflöd inom tech.

SVT:s nya #1procentklubben hade en fråga om att rymma från Svartsjöanstalten, med ett kodat meddelande om “onsdag”. Idag är det onsdag och någon har rymt från… Svartsjöanstalten. Slump? Jag tror inte det. 😅

Amazing we get so much out of the competition between Open AI and Anthropic right now, what a time to be alive. Name a better rivalry in tech right now. I’ll wait.

Been hearing a lot of great things about @EffectTS_, is it worth the +90k LOC for future maintainability and agent speed? I let GPT 5.6. Ultra work for approximately 5 hours yesterday (burned a banked reset). Then later Astra dropped, so I had it review and fix things.

Hello, I've just created a tracker for Swedish trains. If that's relevant for you. Can be seen here: sweden-rail-day.vercel.app/ One caveat which you might already know (or not?) is that this data is based on PLANNED routes, and times. It's not tracked the same way as an airplane..

I got inspired and built this for Sweden! 🇸🇪

🙌🫡

I’m 100% sure I saw a post about a rust version of Postgres just the other day on Hacker News…

This was very confusing…

When a single community note kills an entire AI startup idea 💀

This obviously depends on how much money they will earn from people paying for Fable 5 API usage after the 7th. If people turn out to be willing to pay more for Fable 5 they will obviously not include it after 7th. They are conducting market check, for business decisions.

Atamayanlara atarlar

Anthropic vibecoded their shitty refund page: xitter.cf/OzCode/status/20658714…

Same here. xitter.cf/OzCode/status/20658714…

I am mainly after stuff that can help groom my codebase and keeping it up to date etc in ways that add value. Not looking for a ”do a code review of the changes in main last x hours” type of automation, or ”check my gmail and do a summary” stuff.

How are devs using Codex Automations/Claude Routines successfully? I have a rather large repo I’m working in solo, and so far the only thing I’ve come up with is that it updates the user facing docs once every 48h… There must be tons of things you can do? #codex #claudecode

I bet! How do you guys use agentic coding for the development process right now? I'm thinking both for developing features obviously, but more curious how you use it for code review and validation? Have you found proper processes that fits your team to ship solid and fast code?

Use the fuck out of the shit you build.

This is the way. Honestly, if you sit in your app/system constantly WHILE developing with agents then you know where shit will break in relation to the changes the agents just made. Manual verification, together with letting agents review + re-review a couple of times. 🔥