Sabitlenmiş Tweet
Narayan Babu
22.6K posts
Narayan Babu
@nyn
VP @Zetasuite, builder by night • Shipping https://t.co/GiTQZjzhMe 11 solo builds in 2025 • Scaled : Truecaller (2018, 150M DAU), Glance (2021, 200M DAU), Iris (2011, 20M DAU)
utopia of optimism Katılım Temmuz 2007
134 Takip Edilen1.6K Takipçiler
Time to promote your product. 🚀
Share that product URL!!
English
I make releases to improve existing features, by simplifying them or removing (parts) of them. But sometimes, you think a lot and add some 'new features'. This release is one such.
English
notchpad 0.8 is here. A lot of polish and delight thrown in. But most importantly, adds on device intelligence and wisdom powered by Apple Intelligence.
Narayan Babu@nyn
I have always scrambled for a quick scratchpad for Mac, so I built this for myself. It is an 'always on', offline note/dicta pad which is non-intrusive and skips the entire window hierarchy. (Sort of like Spotlight). Check out at notchpad.org
English
that's what I am doing for the 'app' itself, but like you called out. In fact in the app itself (about dialog) when I 'programmatically' pull the icon, it still shows the glow.
Pulling the app icon from NSApplication.shared.applicationIconImage, which is the runtime app icon that macOS resolves from your Assets.xcassets AppIcon set.
English
That’s exactly what I’m talking about, you don’t need to export anything to PNG. You create the icon, then just hit “Save”, move that single IC project file into your Xcode project alongside your main files, and in AppIcon you simply specify the name of that icon file. That’s all you need.
English
✨ I started enjoying it a lot when I understood how Apple’s Icon Composer works. Now I build all icons there and will never go back to those sets of dozens of sizes
English
@dmitriychuta Case in point -> notchpad.org I am generally happy about the icon itself (how it has turned up), but the border glow is part not so much.
English
@dmitriychuta But if I’m exporting the icon for the web (png) it always renders with the glow. I haven’t checked how it shows in mac (mac apps), but I think it renders the glow as well.
English
@T_Zahil Another problem with 'ship fast' is that, you end up prioritising 'low hanging fruits' (both in terms of features & products) which means you never build anything which is a 'step jump'.
I am equating ship fast === fast iterations as well here.
English
Unpopular opinion: ship fast doesn’t work anymore
Building a lot of products you give up on after a few weeks is not a good idea
Every product you abandon, you’re loosing trust of your customers.
I personally wont buy your products if you keep building new ones all the time
English
@raphaelsalaja Wow. This is super epic. Till now, I have always cloned the repo onto my active repo and asked claude/codex to refer to the content. This is massive value.
Also, amazing 'tool' you have built. ❤️
A lot of the experience of notchpad.org was crafted with ui-wiki
English
i've distilled everything i've written on userinterface.wiki into a single skill file.
119 rules across 11 categories across animations, timings, ux laws, typography, audio, and more.
npx skills add raphaelsalaja/userinterface-wiki
English
@benjitaylor this looks super fidgety fun. Did you finally decide to ship it ?
English
Playing with this fun and completely unnecessary approach to 'severing' OpenClaw connections in Readout. TBD whether or not this actually ships. I think it may have to though...
English
It was not built, but 'crafted' from every sense. I had so much fun, polishing and perfecting every element of this product. Try it out, it bring a smile to you when you use it. Gonna be always free (and delightful) to use.
English
I have always scrambled for a quick scratchpad for Mac, so I built this for myself. It is an 'always on', offline note/dicta pad which is non-intrusive and skips the entire window hierarchy. (Sort of like Spotlight).
Check out at notchpad.org
English
weekend project. do a reverse engineer and rewrite in rust of NES's Contra purely using the latest models.
- blown away by how far I have gotten without knowing the first thing about reverse engineering. the models are *very* good at reverse engineering .
- I still needed to give input when it was clear a path was a dead end or to try an obvious alternative. E.g "search the web for anyone else who has reverse engineered the sound databanks"|
- biggest win was when I could set up an automated loop. Setting up "take a screenshot from the emulator and compare it to the current build and reduce the difference" run overnight made me tremendous progress.
English
Super surprised to see there are no 'truly native' clean and usable 'markdown' Editor component for Swift. Most of the one's are based on webview ( codemirror, prosemirror) which adds latency and also extra entitlements.
English
@steipete But why? You can in parallel run any number of projects/agents in codex? Unless we need to use multiple codex 'accounts'?
English
codex app needs multi-window, but until then, copying the binary totally works
English
I will take it a bit further. I think taste is when you care about the 'product', we keep talking about 'artists' and 'craft'. What separates them from merchants? Artists don't create art with the intention of a 'business outcome'. Taste is about caring **only about** the product, once the product is lovable and magical, it "could" become a business.
Taste is about slowing down, taking your time to look at the product, reflect and how minute details can be improved. It is hard, exhausting and time consuming (for a uncertain 'business' outcome), but deeply gratifying.
Vlad Stan 🔭@vladstan
Everyone says taste is the moat in the AI era. Paul says taste is just caring enough to be honest. I'd go one level deeper: the moat is deep understanding of the problem. A beginner sees an unexpected chess move and calls it brilliant. A grandmaster just smiles. Same move. Completely different understanding of what it means. Same with taste. Becoming a grandmaster takes years, and years alone isn't even enough. You can spend a decade playing chess and never get there. It takes obsession, feedback, pattern recognition built over thousands of real positions. So if you want a moat, build deep understanding of your problem and your customer. Talk to them obsessively. Live in their world. Study what they actually do, not what they say. Ideally, you are that customer. It will take longer than you want. But that's exactly what makes it a moat.
English