This seems to be the generally agreed upon direction defense companies are going, but a couple architectural concerns come to mind regarding this "Manned-Unmanned-Teaming" approach:
- Even if the XQ-58 has a low radar cross section, a swarm of four drones flying in formation with a non-stealthy Eurofighter significantly increases the aggregate probability of detection. Unless these drones are performing active electronic countermeasures or "blinking" to spoof radar returns, they’re essentially a giant "here we are" sign for any modern radar. I wonder if they've compensated via the flight software to manage formation geometry to minimize the group's total observable signature?
- Anti-air systems will prioritize the command aircraft (the Eurofighter) immediately. If the C2 link is severed (kinetic kill, high-power jamming) what is the state-machine logic for the subordinates? Do they revert to a fail-passive (return to base) or -active (continue last assigned strike) mode? Without a human-in-the-loop, rules of engagement issues are abound. (I'm not even accounting for the fact that the drones probably rely on calculations from the command craft, so edge-computing will factor in as well.)
- They're calling these "attritable," but at $4M a pop plus the cost of the sensors, they aren't exactly disposable. Is the cost-per-kill for an adversary’s interceptor missile actually higher than the cost of the drone it's hitting?
(1) Aircraft rarely fly in anything close to formation in combat - large gaps are the norm (1-10 miles), and one would think that increased distance is something that could be exploited by an unmanned platform (able to take more risk, etc.)
(2) Remains to be seen.
(3) Individual Patriot missiles are around that price point, with S300/S400 anywhere from 500k-2M depending on capability. One would think that cost-per-kill would be favorable considering the increased capability granted.
At 10-mile intervals you're maintaining a high-bandwidth, low-latency mesh network in a contested electronic environment. If the command aircraft is 10 miles away and the enemy is jamming the link, the drone is going to be making split-second (potentially) lethal decisions without the pilot.
You're right about them both costing about the same, so the real leverage only comes if these drones can stay out of the engagement envelope while sending cheaper submunitions (likely using something like these Ragnaroks (~$150k) https://www.kratosdefense.com/newsroom/kratos-unveils-revolu...) to do the actual baiting.
> high-bandwidth, low-latency mesh network in a contested electronic environment.
Hard to win at jamming, when you're further away and the opponents are frequency agile.
1. They can use directionality more effectively to their advantage
2. Inverse square law works against you (unlike e.g. jamming GPS where it works for you).
3. They can be frequency agile, strongly rejecting everything outside of the 20MHz slice they're using "right now"-- and have choices of hundreds of those slices.
Fighters already have radars that they expect to "win" with despite that being inverse fourth power, a longer range, and countermeasures. They can send communications-ish signals anywhere over a couple GHz span up near X-band. Peak EIRP that they put out isn't measured in kilowatts, but tens of megawatts.
My concern is less total link loss than what happens under degraded or intermittent connectivity. If the wingman still depends on the manned aircraft for tasking or weapons authority, then the interesting question is how it behaves when the link is noisy rather than gone.
Stealth is less effective against long range radar, stealth is more effective closer in against targeting radars.
When you're high up you can have pretty long 'line of sight' so it's not unreasonable that these could fly way way ahead. 100 miles and way more is not unreasonable.
You basically get 'double standoff'.
I can see this as being almost as effective as manned stealth and if they are cost effective they could very plausibly defeat f22 scenarios.
Once you add in the fact that risk is completely different (no human), then payload, manoeuvrability, g-force recovery safety, all that goes out the window and you have something very crazy.
3 typhoons with 2-3 'suicidal AI wingmen' each way out ahead is going to dust them up pretty good at minimum. It's really hard to say for sure obviously it depends on all the other context as well.
That may be true, but it seems to strengthen the case for moving the human out of the forward cockpit rather than keeping them there.
If the unmanned aircraft are the ones flying far ahead, taking the risk, and extending the standoff envelope, then why is the human still sitting in the forward fighter rather than supervising from a safer node further back?
At that point it seems like the architecture is optimizing for tactical latency and current doctrine, not necessarily for the cleanest end-state.
at 10 miles, the data link cannot be jammed. and it won't be observed, either.
military is very good at this 'mesh networking' thing. L16 is 40 years old at this point, I expect they have something much better.
Looks like they're using some new variant of branding font for this. Inspect Element shows it as SF Pro Display, but it's actually just being masked over with an image
They don't call the base model 'iPad Neo.' They just call it iPad (https://www.apple.com/ipad-11/). It's the same market segment and even uses the exact same color palette.
They also just established that 'e' is the designator for budget model (https://www.apple.com/iphone-17e/) so best guess is they thought 'MacBook e' looked strange so instead it's 'nEo'. And don't forget the 2004 eMac.
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.
I took a few shots at building desktop apps with Tauri, Wails and Electron using Claude Code, and the results were not very good at all. In fact, they were by far the worst results I've gotten with the tool. I can easily clone one of my boilerplate repos in Rails, or Django and prompt away, and the results are consistently good, as in, functional MVP in a few hours. This was never the case for the desktop tools I mentioned.
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 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.
Well, to be fair, I do have an experience working on a Windows Forms app from scratch. App connects to a very specific scanner via customs drivers and makes use of a remote API for data tasks. The app works, it's stable, but I'm not going to lie, AI assisted coding for this particular stack does require a very large amount of nurturing, it is just not the same experience you get with web apps. Nevertheless, it did it.
Makes sense. There's plenty of freely available code and data online for using web tech. Any number of free online bootcamps spawned in the mid 2010s are full of "Become a React developer in 6 months" type of content.
Native, especially on Windows and macOS, have been the domain of proprietary apps there's not much code outside of tutorials online to train a model on outside of official documentation.
I made a couple of small menu bar utilities for mac using Gemini, and it was OK at best. Kept wanting to use deprecated APIs, but with a lot of handholding I got them to work.
Would be neat to see Apple put out their own model specifically for Swift/SwiftUI
I have been seeing more and more native desktop apps in the past few months (octarine for instance), but most of them would've honestly been better off as web-apps, or at least a polished electron app.
> seems to be hostile to any AI generated code at all for now.
Because the majority of vibe-coded apps are low effort.
Yup, I know it's been around for longer, probably wasn't the best example. But it's just the first native app I've thought of and with how much it's been changing, it constantly feels new.
I do like most of it, but the pace of upgrades is a bit too fast for me compared to obsidian, which feels more stable for now. There's also parts of the obsidian editor (the plain-text view, I never use preview mode) that just feels better than every other notes app I've tried so far. Although obsidian as a whole is something I'm also trying to move off of.
Love the polish of octarine though. Has the revenue been decent so far?
Ah the fast pace of updates is because I quit my startup job to go full time on this since last September! So it's a day job for me now, which means I don't need to only spend a few hours per weekend, and thus can get to my backlog faster!
As for revenue, it did give me enough confidence to quit my day job (was pretty well paid for my country), and Octarine since the past 3 months, has exceeded that as well :)
I wanna say reddit? But it's a mix of things, some users come from chatgpt, some from searching for competitors on google.
I don't do marketing (I do a post on reddit once in 3 months or so for updates, but it doesn't get that much traction). Feel like it's word of mouth. Some of the early users told more people they knew, and they did the same.
Now a ton of customers bring the name up in their reddit threads (like you did here), and that's generally it.
I'd love for conversions to be higher compared to the install count, but it's still healthy for an indie project with a relatively higher price point (people are too used to free, or $19 products).
I’ve had a totally different experience. I’ve coded 3 different Tauri apps and 1 Wails app with Claude Code and it was some of the easiest work I’ve done with AI assisted coding. That said, the local features that Rust is handling in the Tauri app is not anything heavy, just moving files around, some regex matching, and some SQLite stuff. All of the headache I had in these apps was the React frontends and Node issues. The Rust features all worked pretty much first try every time.
They say they're targeting Mac only for now, so it could be native code, or they could just have not tested/refined their prompt for other platforms yet.
> 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.
I've tried using Codex and ChatGPT while working on a small SwiftUI app. It's not very good when it comes to newer APIs and features - I imagine due to lack of data about these things. Very often it would rather push something AppKit-based instead of SwiftUI.
It works, but feels really janky and messy.
I had one very annoying bug with file export API where extra view on export window would appear with a delay. No matter what I tried it didn't manage to fix it. Instead it would go on to try and completely rewrite whole file export class in various ways... which still didn't work as it claimed it would. Ended up fixing it manually by caching instance view locally.
> 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.
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.
> 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.
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.
> 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.
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.
I think this is exactly the crux: there are two different UX targets that get conflated.
In operator/supervisor mode (interactive CLI), you need high-signal observability while it’s running so you can abort or re-scope when it’s reading the wrong area or compounding assumptions. In batch/autonomous mode (headless / “run overnight”), you don’t need a live scrollback feed, but you still need a complete trace for audit/debug after the fact.
Collapsing file paths into counters is a batch optimization leaking into operator mode. The fix isn’t “verbose vs not” so much as separating channels: keep a small status line/spine (phase, current target, last tool call), keep an event-level trace (file paths / commands / searches) that’s persisted and greppable, and keep a truly-verbose mode for people who want every hook/subagent detail.
I'll look into what's going on with some of the other browsers.
To clarify, the game actually runs a quick validation when the timer runs out to check if your word is valid. If it is, the ball returns automatically—so you don't have to hit Enter or Space, but doing so early gives you a speed bonus.
As for getting rid of Enter/Space entirely, auto-submitting can be tricky with compound words (e.g., should it submit 'REGULAR' or wait for 'REGULARLY'?).
I was suspicious when I first coded this that it needed a better way to introduce the rules. I've gone ahead and buffed the HINT so that time doesn't resume until you get a chance to type out your first word. I also added a background hint on the return phase instead of just leaving it blank.
Another comment mentioned that the instructions that I put on itch.io made it clearer (and forgot to post on HN, whoops!) so I've pasted them below.
-------------------
HOW TO PLAY
1. Type HIT to serve
2. Type your opponent's word to line up your character with the ball, then type your word to send a volley back
3. Submit your word before the time runs out. The faster you submit your word, the faster your hit!
- Even if the XQ-58 has a low radar cross section, a swarm of four drones flying in formation with a non-stealthy Eurofighter significantly increases the aggregate probability of detection. Unless these drones are performing active electronic countermeasures or "blinking" to spoof radar returns, they’re essentially a giant "here we are" sign for any modern radar. I wonder if they've compensated via the flight software to manage formation geometry to minimize the group's total observable signature?
- Anti-air systems will prioritize the command aircraft (the Eurofighter) immediately. If the C2 link is severed (kinetic kill, high-power jamming) what is the state-machine logic for the subordinates? Do they revert to a fail-passive (return to base) or -active (continue last assigned strike) mode? Without a human-in-the-loop, rules of engagement issues are abound. (I'm not even accounting for the fact that the drones probably rely on calculations from the command craft, so edge-computing will factor in as well.)
- They're calling these "attritable," but at $4M a pop plus the cost of the sensors, they aren't exactly disposable. Is the cost-per-kill for an adversary’s interceptor missile actually higher than the cost of the drone it's hitting?
reply