A bit part of this that people are not understanding is that a major part of the author's success is due to the fact that he clearly does not care at all how anything is implemented, mostly because he doesn't need to.
You get way farther when you have the AI drop in Tailwind templates or Shadcn for you and then just let it use those components. There is so much software outside that web domain though.
A lot of people just stop working on their AI projects because they don't realize how much work it's going to take to get the AI to do exactly what they want in the way that they want, and that it's basically going to be either you accept some sort of randomized variant of what you're thinking of, or you get a thing that doesn't work at all.
> For me it’s just a huge force multiplier, maybe 10-20x of my ability to deliver with my own knowledge and skills on a web dev basis.
I can tell you that this claim is where a lot of engineers are getting hung up. People keep saying that they are 10, 20 and sometimes even 100x more productive but it's this hyperbole that is harming that building style more than anything.
If you anyone could get 10 to 20 years worth of work done in 1 year, it would be so obvious that you wouldn't even have to tell anyone. Everyone would just see how much work you got done and be like "How did you do 2 decades worth of work this year?!"
It's also worse than that. It uses the Raycast logo directly in the launcher itself. Which is odd because just above this, OP says:
"I've actually thought about that before; I've tried to be extremely clear in the project's README with a disclaimer that this is a non-commercial hobby project and is not affiliated with the official Raycast team in any way."
Clearly a bright kid, but that's quite a fumble. Among my ideas for being extremely clear about not being affiliated with Raycast I would have to say using their name and using their logo together would be the worst way to communicate that.
> It can reveal “unknown unknowns”. Often, prototypes uncover things I couldn’t have anticipated.
This is the exact opposite of my experience. Every time I am playing around with something, I feel like I'm experiencing all of its good and none of its bad ... a honeymoon phase if you will.
It's not until I need to cover edge cases and prevent all invalid state and display helpful error messages to the user, and eliminate any potential side effects that I discover the "unknown unknowns".
I think you're talking about unknown unknowns in the tool/framework/library. I think the author is talking about unknown unknowns in the problem space.
I was talking about both. Sometimes even in a problem space time constraints demand that you utilize something off the shelf (whether you use part of it or build on top of a custom version of it).
Tools aside, I think everyone who has 10+ years can think of a time they had a prototype go well in a new problem space only to realize during the real implementation that there were still multiple unknown unknowns.
Yeah, typically when you start thinking something through and actually implementing stuff you can notice that some important part of the behaviour is missing and it might also be something that means that the project is no longer feasible
Yes. I wanted to warn about a rough draft being too rough. There are corners one shouldn't cut because this is where the actual problems are. I guess that rally pilots do their recon at a sustained pace, otherwise they might not realize that e.g. the bump there before the corner is vicious.
Ye it is something like how making tools just you yourself use is so smooth. Like, they can be full of holes and be a swaying house of cards in general but you still can use them sucessfully.
Optifye is positioned hard as a way to make it easier for management to squeeze every drop of blood out of workers, it's pretty fucked to try to use the real issue of women's safety as a defense for this. If this had anything to do with women's safety, that video and the website would have been completely different.
I could probably be a pretty damn good cook, perhaps even a sous chef. But there is so so much I want to do that is not cooking. Is it really so difficult for CEOs and devrels of AI companies to imagine that some people don't want to spend their time making software, even if it is accessible?
If we stick with the comparison to cooking, I think AI is not like a cookbook, but like an air fryer.
A cookbook lets you cook like Thomas Keller or Eric Ripert or whoever because you get their exact recipes, even when those recipes have 18 steps and take 7 hours.
An air fryer lets anyone make something acceptably tasty in 20 minutes and with zero technique.
Imo this is what AI is doing for software building. It lets almost anyone build something that accomplishes a job. But we don't have to expect it to have scalable architecture, beautiful UI or follow security best practices.
I think part of the problem is that most software engineers rightfully care about these things for their jobs. But maybe they don't matter if you're building "home-cooked software".
The same way you can read cookbooks from 3 michelin star chefs and wonder if anyone will ever follow the recipes in them (check out Marc Pierre White recipes for an example).
Cooking (manually) takes skill (let's say that's a one-time thing you need to have acquired in the past) and then it needs time (repeatedly).
The outcome (food) is a temporary benefit.
Software meanwhile, in general, is written once in order to solve a problem at all, or make a task easier/more repeatable. The software is the cook, so to speak.
A more apt comparison would be to craft a cooking tool that makes your work easier, or possible at all. (e.g a pot)
I reread that part of the post and this makes me disagree even more (not with you):
> The more that build software, the more who will appreciate using great software. Cooking at home makes you appreciate going out to eat.
Like.. no, I don't appreciate great cooking in restaurants.
I'm eating out because I hate cooking and I don't know how to cook certain things and with my tools I am not able to cook certain things. (e.g. good pizza). But the stuff I like cooking (and can do) is not worse than the same dish in a restaurant, sometimes I even prefer it. Same for software.
But I guess that's the thing. If I'm a professional software developer it would be kinda weird if I hated using my own home-made software? So maybe this is actually more true for people who just need a thing but can't do it without (AI) help?
How much is this saying when just a couple decades ago a 9 year old could imagine that a broomstick was a sword and that they were fighting knights on the battlefield?
A child can fill in tons of gaps with their animation (see: Rugrats). We're talking about "is AI good enough to enable a 100% non-technical user able to create software on the fly without the non-technical user losing patience?" This is the discussion.
I understand where this guy was coming from with the comment about Tauri being the same type of thing, but when you really take a look at Tauri, it seems to really hold up. There are some Tauri apps that have performance that is easily on par with native.
The only thing is that Tauri apps seem to be quite easy for the developer to botch and end up with performance problems. One of the worst performing apps I ever used in my life was a Tauri app.
> The only thing is that Tauri apps seem to be quite easy for the developer to botch and end up with performance problems. One of the worst performing apps I ever used in my life was a Tauri app.
I'd say that's generally true with web-based UIs. It's possible to create web UIs that perform fantastically. But you have to know the platform and be disciplined, and it's easy to mess it up.
It's actually not Electron, it appears to be Chromium Embedded Framework (CEF), which is what Spotify desktop app uses. I'm speaking completely separately from OP's product:
CEF was supposed to be more efficient than Electron, but I think that framework heavily missed the mark. I'd be interested to know why the dev chose CEF instead.
I actually use an Electron-like framework to make ClipBook, but it's not Electron and not CEF. It's Electron for C++ developers. It's called Molybden SDK. I'm one of the creators of this SDK. I decided to use it, because I know it really well and it allows me to write using C++ and closely work with the operating system. It's very important when you build a software like ClipBook.
I very much dislike the tweet that OP uses on the Magic UI site that's directed at Linear:
"Companies spend $30,000 and several weeks to get designs like this"
Yes. Because being a first or early mover is expensive. You commoditizing their style after they've put in all the brain work to create or build it isn't the same thing.
Also, isn't the point of design engineering to be capable of coming up with relatively original and innovative UI?
> isn't the point of design engineering to be capable of coming up with relatively original and innovative UI?
that's definitely why I got into it years ago, but it seems like it has lost it's meaning and been simplified into "engineer that likes design" or "engineer that wants to be good at design"
It definitely is sad to see people not want to deviate even a little bit. I've seen at least 3 instances on Twitter where someone launched a product and they had the same landing page as another popular app even where certain components don't make sense.
A classic example is the rotating globe that you can click and drag. That was usually meant for showing sales happening across the world, but now people are just throwing it in a card with the header: "Innovative new features".
Like there's just no thought put into anything and what bothers me is how popular low-effort is becoming in tech. Startling.
You get way farther when you have the AI drop in Tailwind templates or Shadcn for you and then just let it use those components. There is so much software outside that web domain though.
A lot of people just stop working on their AI projects because they don't realize how much work it's going to take to get the AI to do exactly what they want in the way that they want, and that it's basically going to be either you accept some sort of randomized variant of what you're thinking of, or you get a thing that doesn't work at all.