Poul H. Hansen

@toddmotto I’m just gonna go in there and stay for like an hour 🫠

@insanicae it's worse just sat at a laptop than out in the sun running in it haha, imo anyway... a nice cold shower after and all is well again

32°C / 89.6 °F heatwave 5k run 🇬🇧 ☀️ sometimes its nice to try new things

@toddmotto Ah yeah. I hear others say it’s not as bad as they imagined. Maybe I’m just using it as an excuse! 😅 Important note: I’m a redhead. So I already burst into flames just thinking about the sun ☀️

@insanicae was enjoyable actually, weirdly cooler than you'd think because you're getting breeze from running haha, though a great tactic was a bottle of water with me and tipping it over my head every few minutes!

@juristr Woop! Noice!

A lot of work went into this one. Check it out! And if you watch the video, LIKE it!! 😜

@daviddalbusco oh that is pretty nice, and ❤️ zod!

Blogged about how I made Server-Sent Events pub/sub type-safe using a singleton wrapper and discriminated union callbacks. daviddalbusco.com/blog/type-safe…

So i'm back to manually writing code for a while and it's fun. I never went 100% agentic, but with current copilot changes our IT department is..... trying to figure it out. So for now it's "ration", prompt wisely or.... use M365? gotcha. I'm fine on manual 🤡

when you prompt the agent in the wrong terminal session and it is completely confused

@juristr Sounds like a nice loop 🤔 was tinkering with something similar a while ago. Albeit not as thought out 😅

@insanicae haha thanks. I’m using the app to test more autonomous AI development by basically building the app itself 😅

Fun to see how other orchestrators keep adding features I have in my small little personal side-project - CLI to fully control the app via an agent (e.g push/complete tasks) - remote execution (run agents on ur mac mini) (👇 no sound, srry, literally just recorded a test-run 😅)

@DanielGlejzner I do something similar. Especially with larger plans, although we usually try and keep it small and focused. But we have it write plans to md files also. It will update its progress as it goes, and it will also record findings. Neat way of catching gotchas or edge cases too!

One slightly old-school thing I still do when running larger AI-driven changes I ask the AI to maintain a report .md file. It documents: - what changed - why it changed - a small code example But the key rule is this: - the AI must update the report after every change Otherwise, once you go through multiple iterations, context is lost and the report becomes useless It doesn’t replace code review It makes it targeted Instead of reverse-engineering intent from a massive diff, I first validate the reasoning, then focus on the risky parts

@juristr Experiencing the exact same thing. At the end of the week I'm so tired of context overload. So been dialing back a little. Ideas still exist, we can move quickly if needed, but not with everything all at once.

One thing AI has made harder for me: focusing and prioritizing. Everything seems at arm's length, easily achievable, so the temptation to parallelize has never been as high. Most of the time this makes you less productive 🤔 Focus on what gets out the door, not WIP volume

@tveskov Sjovt 🤔 i min boble oplever jeg større tilstrømning til f.eks GPT-5.3-Codex og 5.4. Også fordi opus er så meget dyrere

Det er ikke heeelt som Mac vs Windows, men:

@juristr GPT-5.4 being snarky with nx :D

@ipwanciu Also a good practice for componetizing list elements. So like `li[my-list-item]` and such. Not only more readable, but semantically correct and accessible

I don't know about you, but this is what I usually do when I create a custom header, section, footer, or similar component in Angular. This is a small trick that makes the DOM look much more semantically correct. #Angular

@JohannesHoppe Oh what a lovely little util!

🅰️ Angular Tip of the Day: Your app talks to multiple APIs - but your interceptors can't tell them apart!? Wrap it. Filter it. Done!

@tveskov Jeg tager 2 🫡 …eller

Altid god spas at maxe en ny MacBook Pro helt ud: apple.com/dk/shop/buy-ma…

@juristr Maybe it’s just human? I feel dumber on weekends too 😂 Jokes aside I hear this a lot across the board. So maybe there is something to it 🤔

I sometimes feel Claude is dumber on weekends 🤔

@MiniMax_AI @openclaw I've used it for secondary bots and testing. It is superb! Only thing for me is that it's not fluent in Danish. Not a deal breaker, just funny seeing the spelling errors

You don’t need a mac mini to escape the permanent underclass. $10 MiniMax coding plan. $5 Railway hobby plan. @openclaw + @MiniMax_AI = 24/7 personal intelligence.

Used subagents for the first time at work. We've had copilot access for a while but recently pushed back for more usage and removal of some very strict security stuff so we could actually test things out. And boy, while they were super simple tasks, so magic :D

Loving @openclaw ! Been part of the discord for what feels like a week? so much inspiration 🤯 Testing things out, mostly. Today my wife asked if we could set up an assistant for her and her business to help her with scheduling and SoMe 🦞!