Key Features
Tech Stack
I don’t pipe up very often, but I visit HN almost every day. Many of the books I read, blogs I frequent, and podcasts I listen to, I found via HN. I think it’s fair to say, if it weren’t for HN, that Bitrig wouldn’t be Bitrig and SwiftUI wouldn’t be SwiftUI.
- How to Win Friends and Influence People — This was really eye opening to me and helped me more productively engage in a large organization like Apple.
- Sam Harris's podcast — I'm always interested to hear his takes on the controversial topics of the day as well as more philosophical and philanthropic ones.
- The Knowledge Project podcast — I don't listen to this as much anymore but a few years ago I was a regular listener.
- Simon Willison’s blog — The highest signal to noise way I've found to stay up to date with developments in AI.
And I echo the thanks. HN has made me a better technical professional in so many ways.
For the most part, I think the look and feel of the apps benefits from SwiftUI baking Apple's design system into the defaults so heavily.
Really cool product, as someone currently attempting to build a somewhat similar internal tool I have an understanding of some of the pain points involved.
Please don't allow yourselves to be bought out by Apple in the way Buddy Build were back in 2018 though! (and then shut down)
Also, side note: magic link or email-based OTP login is by far my least favorite method of login, especially for a phone app. It's cumbersome, annoying, and completely unnecessary now that passkeys exist. Barring that I'd still rather use email/pw login any day of the week.
We've definitely been hearing from users who don't like OTP, we'll try to get additional login options added soon.
Now that I’ve gotten to play around with the app I have to say I’m impressed with how smooth it feels and how well the Swift interpreter works (at least on the basic prompts I’ve tried). One suggestion: allow toggling between an “edit” mode and a “view” mode. The latter would full screen the app and hide the interface for follow up prompts, etc. that would make it easier to approximate how the app will behave in the wild.
I could see this eventually turning into sort of a social platform for mini-apps, by the way! Where users don’t even necessarily need to publish to the App Store if they’re just creating something for themselves and a few friends.
I really like the idea of a platform for people sharing mini-apps! It would be great if people could make lightweight specialized apps for different niches and share them with less friction.
Versus for email OTP you have to switch to your email client, find the email, perhaps refresh a few times waiting for it to come in, click into the email once you see it, highlight the OTP, click copy, then switch back to the app and finally paste. I don’t understand how anybody could possibly prefer it.
In the meantime, one thing that's worked well for us is building a component/screen/etc as a standalone piece in bitrig, and then bringing it into a larger project from your computer later. You can easily export the code you make in bitrig for those kinds of workflows.
Really large and complex apps will probably require improvements to both Bitrig and the underlying models. However, both those are coming :)
Are you imagining you'd be editing the code with the software keyboard or a paired hardware keyboard?
My workflow often goes like this: got an idea, let me quickly test it on the go, like while walking down the street, working on unrelated stuff or just chilling. Open up Pythonista, write some code, run it, tweak it, etc. If a piece of code becomes important enough, I copy it to the computer and continue working on it there.
Much slower (minutes) with Xcode Cloud and a git repo, if I can git commit from my iPhone, I can edit app code and commit, watch Xcode Cloud Build see my commit, watch it build with debug logs, and when it works it shows up in TestFlight. Very much not offline, and minutes long loop.
Keep crushing!
Great work and I hope to become a paying customer soon!
I think most of that would work pretty well for making SwiftUI desktop apps too.
- If Bitrig can do things like let your app have editable data, or make rest api calls, etc. it would be nice to see a little more in the demo video.
- Sounds like you don’t have an expensive mic/studio for recording, but luckily this is free and can work wonders for your audio: https://podcast.adobe.com/en/enhance
Best of luck with the launch.
Thanks for the tip about the mic. Fun fact: it's me in the video, but it's Jacob doing the narration :P
Entered my my prompt… It did a bunch of stuff - apparently initially focused on making me an app icon which I don’t really care about.
I moved away from the app, went back and everything had gone and my prompt was just sitting there by itself.
I said “please proceed” and it opened a new chat, did the same thing. It was doing stuff, I moved to something else, went back, everything vanished.
So that’s two of my five free messages zapped. And all of my patience.
Moving in and out of the project while messages are sending is a more fragile area of the app and we have some bugs to fix there.
What happens if you go over 100 messages/month?
I just burned my 5 free messages to get a simple toggle button working that just says "win" (with animated fireworks!) and "lose". I'm sure I'm not an efficient prompter, but it seems I'd knock out 100 messages easily in an afternoon, which looks to be the monthly limit at $20/mo.
(This is coming from someone who has no idea how expensive it would be to 'vibe code' using something like Claude ... so it may be an entirely unfair assumption that you could chat with this 'unlimited' for $20/mo ... that's what I have in my head as 'reasonable' only because that's what I pay for Gemini or ChatGPT and, for all intents and purposes, it feels 'unlimited'.)
We're still in the learning phase, and are going to adjust the plans based on exactly the kind of considerations you're raising
There's a lot of inspiration we could (and probably will) draw from other products (e.g. Cline's Plan vs Act modes) to build what you're describing.
It can also be really fun and productive to just rapidly iterate on what the AI gives back, without having to take the time to describe it all up front. Sometimes this approach can lead to a new directions you might not have thought of.
With your sunscreen example, I should be able to just ask Siri to do exactly this, and it could tap me in my pocket, show me a custom UI to log periodically and then disappear.
Not sure on limitations of APIs on iOS but definitely feels like there is space for a better voice assistant, just like how Raycast have created a better Spotlight on Mac.
Looks very cool so far!
Our background is more in making developer tools, so bitrig was the first product we were excited about, but I think there's a lot of overlap with dynamic UI generation, and we'd love to explore that too.
That said, we think this same approach could work really well on Android, and it’d be interesting to expand there too.
Seems like a reasonable poll question: Which is harder: web app development or native app development?
I was telling a younger colleague the other day how back in the day, I was "competing" with a peer at a company doing nuclear design software (Siemens Power Corp at the time). I was doing it in VisualWorks Smalltalk, he was doing his in straight C and X Windows. Competition aside, both of us built multi window complex graphical design tools that front ended heavy duty simulators, managed various batch jobs, tons of IO, parsing, and lots of color graphic drawing (mine was for fuel assembly (pin and axial enrichments) design, and his was for core layout (where the assemblies go during each reload)). The graphics were simpler (no antialiasing, any alpha we did with clever dithering tricks, not much in the way of animations, though we did each roll some). Both even stored /retrieved into a sort of database (a homegrown data versioning system). But... it was just one of each us for each "app".
Sadly, all Ux development seems orders of more magnitude complicated and beuqacratic now days. Some things are easier and vastly simpler than they used to be, and yet, the sum result is more difficult. I'm not exactly sure why.
Also, I have been taking the code and advice from Bitrig into Cursor, because it solves problems Cursor struggles with no matter what model it calls.
Your work is empowering, but more than that, it is inspiring, because I can now start to envisage what computer human interaction will look like in the near future.
You have taken away a layer of complexity to creation and that is to be applauded. I think your work is a landmark. I think Bitrig will unleash a new era of creative expression. I wish that I was part of your team.
Could you send us feedback from your project? (Go to the ... menu, then Send Feedback.) If you're not able to use the UI from within your project you can long-press on it from the main screen and choose Recovery Mode, and then enter the project with the interpreter disabled.
Since I don’t have that app on my phone I’m met with an error message and cannot know what address of yours I need to use to send a feedback.
Btw my feedback is that I tried to create an app which needs camera access. The prompt and code creation worked. However the app does nothing since I believe the permission for camera access are not handled on bitrig
I'll take a look at camera access and see what's going wrong there.
I'll likely subscribe to your service.
For distribution strategy you could experiment with letting people create simple apps for free and put a perpetual ad for your service ad the bottom of the app.
Submitted request for WebKit to be added.
In the app I asked it to make, I requested that it use a web view to log into YouTube. The generated app opens a mostly blank screen with no web view, and from what your app told me I think it’s because it’s missing WebKit.
WebKit as a whole might be big, but maybe some subset of it to support web views and get cookies from the web view could be supported?
If we hit issues we can try to pull in just a subset, but one nice thing about Swift is the frameworks have .swiftinterface files which list all the APIs in them, so we’ve generally been able to pull in whole frameworks at a time.
We also have a discord here: https://discord.com/invite/C7ndWKKq2S
So don't paste any tokens or keys into these AI tools.
We aren't doing anything shady or unexpected with the data given the server-based model – projects are stored on the Bitrig server, but we aren't selling the data, using it for training, etc.
We may use the data to improve Bitrig in the future (e.g. by gathering information on the popularity of certain APIs to prioritize bug fixes and new features).
All AI requests use Claude, which of course sends some data to Anthropic.
I pasted your URL into my app and it came up with:
- Non-developers want to create mobile apps without coding. - Users are willing to pay for an AI-powered mobile app builder. - AI can generate functional apps using only a conversational interface.
You can get a compiled build of your app from bitrig: go to Distribute App which will build your app on our server and then upload it to TestFlight for you (this requires a paid Apple Developer Account). This lets you install the compiled app on your Home Screen. You could also always export the code if you want to build it yourself.
I think there's also work when need to do in the system prompt to guide the model away from all of the placeholder data (or at least so that when it adds it there's a better affordance to transform it to real data).
https://play.google.com/store/apps/details?id=com.tyron.code
And well, the app the post is about, for iOS.
I may not like it but I understand the services that do this.
I've been working on a project and sometimes when I'm away from my laptop I have ideas I want to try in SwiftUI but only have my phone on me. What I end up doing is just writing my idea in text and then waiting till I'm back at my desk to give it a shot. Being able to spin up a quick prototype and preview it on the go could be cool.
I don't think my dog would enjoy me lugging my laptop on our walks but my phone is no big deal ;).
You can export your entire project as a single swift file from the share sheet. This makes it pretty straightforward to import the code into an Xcode project when you’re back at your Mac
“What to do in Zagreb with a family?” will build a personalized map & content app tailored to everything the AI knows about your preferences.
AI services will want to own this, but the current OS makers have all the incentives to not let them, so I expected OpenAI hardware to be this - a smartphone with launcher basically replaced by a prompt.
Until then, you might sell Bitrig to them as a way to wrestle this approach onto competing platforms, wrapped inside their app?
As an indie app maker (Weathergraph - a beautiful hourly forecast visualization), I should probably include the Ralph “I’m in danger” gif :)
Also, how many years do you think it will take for AI to reach that level of creating a bugless app that won't need a real developer to look at not even a single time?
Will users stay around for hours for the app to be built and deployed after they ask their question?
Of course - so the AI service can cache these, detect automatically most frequent use cases, and pre-build higher quality apps or building blocks.
What hundreds of thousands of people are doing manually now.
Anyway, congrats on the launch!!
I'm not really sure if I've hit a limit or if something else is going on.
I noticed that all the swift code it writes is in a single file - as opposed to making it more modular. that way it’ll be easier to read the code/make edits. Is this in the works or am I understanding this incorrectly (I don’t have any swift experience)?
Going to try this now
-Chris
I do not own an iPhone and I do not use swift. I just want to say congrats on the launch, this looks and sounds really cool. And watching your interactions in this thread, you look like you'll do really well!
Not affiliated with Hacker News or Y Combinator. We simply enrich the public API with analytics.