How Dactyl Works

jboss10 | 12 hours ago

Sad this isn't about keyboards.

kjloveless | 11 hours ago

not exactly keyboards, but a lot closer then Deno's dacytl

https://github.com/kjloveless/dacytl

funny enough i wrote this years ago using Deno!

geooff_ | 12 hours ago

Their homepage really makes me think that this is actually just an attempt to speedrun cease and desists.

gavinray | 12 hours ago

Interesting tidbit about the author of this blogpost:

Many years ago, on the Deno Discord, this person was posting some really cool PR's they had done to contribute to core.

I checked out their profile, and it turned out to be a 13 year old child (or some similarly astonishing young age)

Neat to see they eventually got picked up by Deno, definitely one of the more talented devs I've ever run across.

ahf8Aithaex7Nai | 12 hours ago

"App development has always been harder than web development for LLMs."

What do they mean by "always"? Am I old? Are there people out there for whom LLMs aren't the new thing?

tancop | 12 hours ago

From the moment LLMs became usable for coding. The web ecosystem is bigger and more unified so there is more high quality training data. Native apps are different on every platform and less likely to be open source.

victorbjorklund | 10 hours ago

I mean copilot has been a thing since 2021 so 5 years.

jakozaur | 12 hours ago

I tried it with a less typical case (the Warsaw, Poland transit app).

I'm impressed that it worked at all and got quite far. Though it is still broken in many ways (random geo locations), the menu bar on top...

The problem is that many web apps are quite simple, while native apps get trickier when it comes to UX. The LLM doesn't work well if it has to interpret animations or cross-correlate map data with locations on a map.

CharlesW | 11 hours ago

Is there anything here that professional product builders might be interested in, or is this just another entry in the AI app builder category?

It's cognitive whiplash to be sent to a blog post that will be interesting to some HN readers, and from there jump to a front page that says, "This not for you, and if we're successful will make make you irrelevant".

oasisbob | 11 hours ago

Technology aside, those are some truly gorgeous and informative graphical examples.

yegle | 10 hours ago

For a moment I thought this is related to https://en.wikipedia.org/wiki/Pentadactyl

(Reading the wiki I'm really glad this is still alive).

krzyzanowskim | 9 hours ago

Very ambitious and fantastic pet project. In my opinion doomed to share the Flutter fate that has similar goal of reimplementing the entire OS UI layer (liquid glass hit flutter very hard for that reason). Technically that "does not scale" after a while, given it's reverse engineering effort. It is far from pixel perfect today, and (my prediction) will not improve much. some aspects are impossible to copy Apple design pixel perfect on the UI level, not to mention on how SwiftUI modifiers may behave. Great work nonetheless

lukevp | 8 hours ago

Running an iOS app on Android is still gonna look like an iOS app. This seems worse than Flutter (reimplement the whole UI) or React Native (abstract on top of native UI.) If we’re ok picking a non-native UI why not just build a Cordova app?

undefined_void | an hour ago

This is an abstract on native UI