Sam

Helping by trying to help myself. Generalist maxxi

[Sam @samboffa]

I know it's being said quite often, but...I feel like it's more important than ever to rely on fundamentals In every field. AI and tech can help, but what if all is gone tomorrow?

Don't build a fundament on sand.

[Sam @samboffa]

Oddly, I find something true in this that rearranged my thinking...

When living standards improve, people’s expectations rise, eventually faster than reality can handle.

Once an amenity like AC becomes widespread, it gets much harder to take away (which can happen by force) without causing societal trouble.

In prosperous countries, maintaining that prosperity gets tougher because people become more sensitive.

I still want more aircon (especially where it matters to be productive), but we should also deliberately train cultural resilience. Singapore is a good example. Comfort is a privilege, not a right.

Otherwise prosperity eventually eats its own foundations. And there is enough historical evidence for that.

[signüll @signulll]

ppl are weirdly disciplined about imagining disaster.

but almost never ask:

what if it all turns out better than you ever imagined?

[joris kanowski @KanowskiJoris]

der europäische weg zu leben

[Sam @samboffa]

4. Iterate step by step

What I do after the initial prompt is iterating with step 1.
It's tempting to do many steps atonce, but if the animation is a lil more complex, codex is going to fk up a few steps. If you end up here, try to split it into smaller steps

[Sam @samboffa]

During the weekend, I might found the best way for me personally, to translate my thoughts into working and smooth transitions in codex 👇

[Sam @samboffa]

3. Use skills!

-> I'm mostly working on react native. The one I'm using frequently is https://t.co/S8GELs9E06. It covers many best practices like reanimated, gestures and skia.

[Sam @samboffa]

@offpaths You should try the cafe at kalm village in chiang mai then, trust ✌️

[Sam @samboffa]

@rsuyoy Positive instructions work better. Negatives force the model to activate the unwanted behavior first, then try to suppress it. Positives point straight at the goal.

That’s why if it’s important I tell the llm to rewrite the prompt

[Sam @samboffa]

@danliu There is no reason to bring Liquid Glass to the web

[Sam @samboffa]

This

[Guinness Chen @guinnesschen]

Bro it’s June 2026. Stop hand editing your prompts. Hold down the dictation button and ramble for 10 minutes. Give the model every fragment, caveat, example, and vibe in your head. It is literally a large language model. If it’s superhuman at anything, it’s reconstructing latent intent from language.

[Sam @samboffa]

@utd_Eman bro looking like straight out of la masia

[Sam @samboffa]

@gabriel1 Also people out here thinking in a bubble. The pie is waaay to big, competition is literally zero in a sense

[Didier Drogba @didierdrogba]

Le haut niveau ne pardonne pas!!! 😤😤

Tous déçus du résultat de ce soir mais optimistes 🙏🏾pour ce troisième match capital pour entrer dans l’histoire.
🧡🤍💚🧡🤍💚🧡🤍💚🧡🤍💚🧡 🤍💚
#AllezLesElephants2026

[Sam @samboffa]

@mionomedev it won't tho, france s going to win the group

[Sam @samboffa]

@robj3d3 it's probably more ROI if you wanna be sure owning the data

[Sam @samboffa]

@__tinygrad__ but makes sense, if you wanna own your data

[Sam @samboffa]

@__tinygrad__ sounds stupid to me

[Sam @samboffa]

@andreasklinger Aldi is superior. They have one of the best own brand stack, quality wise