Not everything is possible yet and sure more complex things need more prompts but you’ll be surprised what Glaze is capable of already. It’s day one…
More seriously, what do you believe your moat is here?
Is it a real synth or license-washed Vital/Surge?
The FAQ was light on technical details. But I am someone keen to read all the technical details :)
I assume there is an extensive set of rails for the agent to tie into. (Compare this to asking Claude to green fields an app. Do you use electron? How are notifications handled? Icons? Permissions?)
It springboard off Raycast’s teams feature so well it actually gives it a real reason to exist. You’re empowering the one systems thinker in the group to export their automations to the rest of the group in a way that’s proven to work: small apps that do one thing. (Big apps get complicated, become full time projects that distract from the task at hand)
Fig tried this but it was just for engineers, the value prop was missing, Glaze seems to get this right.
Very nicely done!
Can you one-shot a raycast alternative with this? This'll be the real test.
(using claude code max by the way)
My first thought was, "So, Replit and ilk?", seems they expected that comparison:
> How is Glaze different from Lovable, Replit, or v0?
> Those tools build for the browser. Glaze builds for your desktop. That means your apps can access your file system, your camera, keyboard shortcuts, menu bar integration, and background processes. Things a web app can’t do. It’s a different category entirely.
Pretty sure modern web apps can do all of those (sans menu bar). (If anything they do background processes better since you can send a very long task off to a server and shut off your computer, come back later and pick up where you left off.)
Also, as others mentioned, this just seems like Claude Code with extra steps, unless they managed to nail some sort of design standard enforcement they feel is better than what most people can get out of it.
The quick publishing is kind of nice, but it immediately made me think it would be more interesting to have a way to quickly remix other people's creations, similar to the Figma Community tab: you can take someone else's work, break it apart to see how it works, then tweak it how you want it.
This looks like a highly specialized tool for desktop that actually works. I watch the demo and I am assuming the apps are actually made with some kind of technology a la Tauri, or Electron, thus making the apps cross-platform.
I don't think we are anywhere near a tool like this for native, but that's a lost battle anyway.
I hope it's not a lost battle, tbh. I was hoping with AI & Vibe Coding we'd see sort of a resurgence of native first desktop apps, but so far it's just all been a continuation of the web app & web tech hegemony.
Maybe not for Windows as their native GUI story is a lost cause now, but for sure macOS and I had hopes of it leading to a renaissance of desktop linux apps in GTK instead of electron, but that (the Linux) community seems to be hostile to any AI generated code at all for now.
> This was never the case for the desktop tools I mentioned.
I'd be curious how well Claude Code works for a native Swift app on macOS, if that's the platform you're on. I've found it extremely good at iOS apps so my guess is it would be equally good at building a native macOS app with the same stack.
Well yeah, isn’t that criticism we’ve had every LLM wrapper for years now? “Show me the prompt!” But that doesn’t mean these types of products are useless.
I think it's fair to say that's a benefit of web apps over native apps in many cases. But for the kind of business app use case they're talking about, it's also a tradeoff. I can imagine a lot of business apps where you don't want to send the data to the server of a Replit etc. and doing all the processing local is a benefit.
My feeling is that it's intended for a less-technical audience than Claude Code.
If you're on Chrome and give them permission, or stuff them into Electron and friends, they can. The workflow isn't as smooth as with native applications, though.
On the other hand, the web browser does protect you from some of the risks this essentially "trust me bro" curl2bash-as-a-service product inherently comes with.
Definitely a good initiative though. I like how coding harnesses do it, showing you the exact command that would run, or running it in a sandbox first.
My experience has been that Opus consistently generates UIs that are genuinely good. As always with anecdata, YMMV.
There's a reason Tailwind Plus has revenue problems right now.
I’m torn about what this likely means for iOS; while I do want to do Raycast-y things in my phone, I’m not sure there’s enough of us to make a business out of it.
I do have prior experience developing for iOS but that was pre-swift.
I'm not using the Xcode integration and so there's still some rough parts where build errors show up in Xcode and I then have to paste them into my Terminal.
When you are used to backend work...it's kinda fun to see an app come to life and run on your phone though.
I imagine that would help the process a lot
We need more of you. Not more electron slop.
Having antigravity is useful because Gemini 3.1 is pretty good at generating UI sugar. Claude 4.6 Opus provides nothing to write home about. Their shadcn looksmaxxing hasn't generalized to writing good desktop UIs.
Raycast's only edge here seems to be the fact that they are obviously very good at Mac app development and probably have impeccable skills/documentation for building them.
Taking a step back, it's pretty clear that Raycast is angling for an Apple acquisition here with this play. If I'm Apple, the reason to buy a product/team like this is a no-brainer.
Raycast recently made a Windows version. So perhaps they aren't as Apple-centric.
No kidding, although I think Apple would only be interested if it uses SwiftUI. (The marketing page doesn't say. Raycast itself uses React + Node for extensions, but its React components render to native widgets.)
I guess they're trying to make more money and Raycast is a bit of a dead-end there.
I get that we've all been running homebrew and npm and so on without auditing every package, but there, there was some semblance of a trust mechanism and distributed review if something went wrong.
Here we have totally unreviewed code running with no trust mechanism in place whatsoever.
No public releases, no audit trail, no versioned releases, no community eyes on anything.
Weird times
I am worried this is the start of them trying to diversify their product offering because revenue has stalled in the core Raycast product and VC demands more returns. I don't want to be jaded, but history teaches me to be. Here's hoping that Raycast itself is still a focus for the company.
I wonder what it is actually building. Tauri apps, maybe?
Similar to how openclaw is exploring a “personal agent” that runs on your computer, this feels like a step toward personal software - tools that live locally, understand the context, and adapt to how we actually work.
Excited to see how this evolves, feels like an interesting direction.
Have it create a swift app, unless extended permissions are needed it can compile withouy going into xcode.
Few simple util apps, disk cleaner, clipboard manager. Worked pretty well.
Had better results than using xcode's built in ai extension.
Design Philosophy Create apps that feel premium, polished, and worthy of being featured on Dribbble's most popular shots. Every pixel matters. White space is your friend. Less is more, but what remains must be perfect.
Visual Design Principles
Color & Theming
• Use sophisticated, limited color palettes (2-3 primary colors maximum) ...even better if the apps are not electron bloated and mac-native.
Even though they portray some of the benefits of this app as unique to desktop apps, they're not (e.g. storing data on device, offline mode etc.).
Am not a hater. Love Raycast. Saw the post and opened the link intrigued what they came up with and was somewhat disappointed tbh. Good luck to them anyway!
For the developers: will Glaze replace Raycast, or are they meant to run concurrently?
Raycast -> Glaze AirTable -> Lovable Competitor Retool -> Lovable Competitor
Even those early in the journey are converging towards coding.
A better replacement to iStat Menus.
A local-only voice to text whisper.cpp transcriber I can globally use while holding ctrl-semicolon.
A menubar app that manages blocky and can easily turn it off or change dns.
A tool like hammerspoon but I configure it via nix-darwin and it has no cruft.
All of these are apps that use 30MB memory and are better than the apps they replace, and I can make changes any time I want. That's far better than using someone else's software and giving it privileged access to my machine.
Also, perhaps the best point is that so much software is junk that is obsoleted by someone with better UX intuitions even if they are vibe-coding it. Being written by hand by an engineer means basically nothing when it comes to "is this a good app?" Which is why product-minded people are the biggest winners in the new AI era.
Vibe apps are different. They do exactly what you want, exactly the way you want it done. No more downloading an app that is mysteriously 180MB and requires watching a youtube video to learn how to make it change your background every 5 minutes to different dog pictures.
Desktop software is nowhere near good enough to consider random usecases "already done". Not that glaze looks particularly special, but there's so many improvements the desktop experience begs for.
An easy to use cross-platform GUI builder for one. Even something as basic as a calendar app doesn't have a clear obvious winner today.
In the DOS days, I would have whipped them up in BASIC. This was standard practice for PC users who were not "software engineers" by trade.
The complication of PCs over the past 30+ years have robbed regular users of this ability.
Tools like this close the gap, and that's awesome.
Whether that's what you want from Raycast as an individual user or not though is a fair question. I was a heavy user, but I recently switched back to Alfred. I love both, and there are things I think Raycast does much better than Alfred, but ultimately I'm not the target market for Raycast and I was feeling for a year or so that I was swimming upstream on my usage of it.
1. macOS and Windows require installation of Xcode and Visual Studio respectively, and if in Apple's case you kinda can install these tools headlessly and choose to install only the "build tools" package, Microsoft's creature is gonna daze and confuse you with a crap-ton of checkboxes and no easy "just install whatever is minimally needed to compile my code" button, and I don't recall if there is way to install build tools on Windows through terminal.
2. what is going to be distributed? source code itself or actual binaries? and what will the security model of Glaze store be? same as extensions, "everything is open-source and undergoes Raycast's and community review"?
3. Glaze is going to come to Windows and Linux, if we trust the Q&A section at the end. what will Glaze build upon? separate frameworks and languages for each platform or something multi-platform [1] like Tauri or Kotlin Multiplatform? or are you going to copy the Raycast extension model - just run Node, expose some platform integration, and parse React render trees through "Glaze Runtime"? I've been working on a bug in Vicinae [2][3], and I've seen this model in action. it's very hard to make it perform well, but all it takes to achieve native look and feel is to just map React render trees to whatever system component OS offers. (in Vicinae's case, it's Qt. bet that it's done with SwiftUI on macOS and WinUI 3 on Windows.)
[1]: there is a difference between "cross-platform" and "multi-platform". "cross-platform" means "I behave equally across platforms and have no awareness of native look and feel" (e.g. Electron, Unity, Flutter), while "multi-platform" means "I can adapt across platforms to the degree you need" (e.g. C/C++, Rust, KMP)
In many ways that's the same as Chrome apps, they have no code, they're just the Chrome binary, so there aren't any code signing issues.
So this is just another Codex/Cursor/Vscode but with different branding to target MacOS apps? Is there any novel features or is it literally just a pure play branding/packaging exercise over the openai/anthropic API to trick normies?
I'm starting to get even more negative on AI coding tools...somehow they provide even less value-add than the trivial AI copywriting apps. At least the writing apps have to contend with building a text editor and workflows around collaboration/approval/publishing, which is actually a real value-add over the raw model API.
But I've yet to find a coding tool with a moat given there's zero workflow differentiation even for large teams -- collaboration, approvals and pushing to prod is all handled by git either way!
that's would be Electron app, but without unneeded bloat
As an interesting counter-proposal to wasting time with this... look for older less popular/downloaded/featureful apps written by people for their own education, edification and enjoyment.
They may not work the way you wish they would, but you can learn a lot from them, be inspired by them, and leave feedback.
That's how you actually encourage more people to get started and continue making their own tools.
I review every diff Claude Code applies and periodically re-review entire projects as a whole. Through this, I've managed to keep architectures fairly principled with future expansions in mind. I recently managed to essentially two-shot an MLX implementation of a working forward pass for a diffusion language model, based on CUDA source code that is not compatible with my machine. There's more work needed before it's anywhere usable in practice, but the fact that the model is now running at all on my machine is a very impressive start.
For that, I had it study the CUDA source code and write a very detailed document with its analysis of exactly how the model is implemented. This document only had one material flaw. Then it studied MLX for a while and spat out a running forward pass based on the flawed document. The output wasn't of sufficient quality, so I had it insert debug prints throughout the whole inference process to see where it was going wrong. It found and fixed the forward pass and the flaw in the document. I needed no domain experience in LLMs or DLMs for this (although I benefit from some minor past contributions to RWKV.cpp).
Another example is that I recently started getting into SwiftUI, and Claude Code is doing a very good job at demonstrating code patterns and pointing me towards APIs that may solve my problems. It also helped me set up things like API clients (which itself, of course, gave me pointers to all the sorts of documentation I'd benefit from reading in full). I reject a very large fraction of its suggestions, I tweak its plans very frequently and I tell it off a lot from things that either are unidiomatic or are objectively terrible hacks. But it is incredibly useful for menial work, for enumerating possibilities, for quickly scaffolding placeholder content, and for demonstrating patterns I haven't learned yet as they apply to my specific situation. For example, Claude Code quickly learned me how to use NSViewRepresentable, whereas in a past project where I didn't use LLMs, I absolutely struggled to embed a Metal view.
But all that is to say that I'm skeptical of solutions that try to have you describe your idea in plain language; that try to insulate you from the code; or that make the lie that you just don't have to worry about it. If you work at all on the kinds of projects I work on, which are chock-full of reverse engineering and an obsessive focus on tightly-controlled design and idiomatic code, I firmly believe that treating the code like a black-box is not the way to do it. I don't know if Glaze truly hides the code, but I don't see any mention of it in the trailer video and that makes me feel a little dismissive.
add this to the bubble theory
ie the annoying way that LLMs interact with users
On the other hand it is kind of the perfect name for Yet Another AI Website Maker (YAAWM?).
> First you said all you want is love and affection / Let me be your angel and I'll be your protection / … / Thought I was a donut, you tried to glaze me
> I ain’t gotta tell you I had a Dunkin' Donuts fetish back in the day. I used to get a dozen donuts every day, man. So it was one of the things that was on my mind
Kids these days are always saying "snaoƨd" and "foʀaṅr"
There is no way a human wrote that page. If Lucas Gray even exists, he should probably reconsider that last image, and his life more generally.