YT YouTube2Text

YouTube2Text → Transcripts → How I Make Apps FEEL Premium (5 examples)

How I Make Apps FEEL Premium (5 examples) — Transcript

by Chris Raroque · 3,371 words · 485 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00The number one question I've been
0:01getting is how do I make my apps feel so
0:03good to use? If you're new here, welcome
0:05to the video. My name is Chris and I
0:06build productivity apps. I've been doing
0:07this for years and I'm finally making
0:09the video that you guys have been asking
0:10for, which is everything I do to make my
0:13apps feel premium. We're going to go
0:14through practical examples in my own
0:16apps that you can apply today. Let's hop
0:19right into it. The first thing that I do
0:20that genuinely makes a pretty big
0:21difference is animations. Now, instead
0:23of talking about this abstractly, let me
0:25show you a real example of something I
0:27just shipped so you can see that there
0:29are levels to animations. I was working
0:31on a new feature for Amy and quick
0:33context, Amy is a calorie tracking app
0:35in the style of Apple Notes. You type
0:36the food that you ate on the left and
0:38the calories magically appear on the
0:39right. The problem with letting users
0:41input whatever they want is the text can
0:43get pretty long. It gets kind of
0:44cluttered so I shipped a feature where
0:46it'll summarize long entries for you. I
0:49asked Claude Code to build it and it did
0:50actually pull it off. Like it
0:51technically works, but watch closely.
0:53[music] If you look closely, the text
0:54just changes instantly. There's like no
0:56feedback at all. And if you blink and
0:58you missed it, you'll have no idea that
0:59the app actually did something.
1:00Technically, the feature is working. It
1:02is summarizing the entries, but the user
1:04can't feel that it's working. And that's
1:06a problem. So that's level one, there's
1:07no feedback, there's no animation, the
1:09user just has to guess what happened. A
1:10pretty junior developer would stop right
1:12there and just ship it as is. But
1:13someone one level up would see this and
1:15say, "Okay, we need to add some
1:16feedback." So let's just add a toast. A
1:18toast is a little pop-up that's going to
1:20appear and say something like the meal
1:22was summarized. That way the user knows
1:23something happened. It does get the job
1:25done and I think it's okay to ship as
1:27is, but we can [music] do better. So the
1:29next level is to add some animation to
1:31convey that the change happened. So I
1:33asked Claude Code, "Can you make an
1:34animation?" And after a bit, it did add
1:37an animation to show the change. Now,
1:39this does get the job done, but I
1:40thought to myself, "Can I do better
1:42here?" And so the next step up [music]
1:44is to take this animation and refine it.
1:47This is what I landed on for the
1:48animation. We have this beautiful
1:49purple-orange gradient that conveys that
1:52something automated or AI or magic ended
1:55up happening. I'm not going to lie to
1:56you and say that I was able to prompt
1:57this in like one or two prompts. This
1:59took some serious work. Like I think it
2:01took 6 hours of prompting. And the
2:03reason it took 6 hours is I tested and
2:05scrapped a lot of different concepts
2:07before landing on this. Even on the
2:09colors alone, I probably tested like 30
2:10plus colors just to land on this purple
2:13orange gradient. Now technically on
2:14paper across the four variations we
2:16have, the feature is exactly the same,
2:19at least on paper. Obviously whether
2:20you're using a toast or no animation or
2:22this beautiful purple orange gradient
2:24animation, the feature feels completely
2:26different. The difference between these
2:28two animations is the effort that I put
2:30into it. One took one prompt and then
2:32the other one took 50 plus prompts and
2:33probably like 6 [music] hours. And it
2:35might not seem like it, but users can
2:36tell the difference in effort. If you
2:38want to build the skill of what makes an
2:39animation good, my favorite resource is
2:42Mobbin. They literally have an animation
2:43section where you can study good
2:45animations and slow them down so you can
2:47kind of study [music] frame by frame.
2:48And that's how you train your eye to
2:50figure out what good looks like. This
2:52video is not sponsored by Mobbin, but I
2:53do have a code in the description that
2:55gets you a discount on their yearly plan
2:56if you want to check them out. The key
2:57takeaway here is not all features are
2:59created equal and not all animations are
3:01created equal. There are levels to this.
3:03Next thing I focus on to make my apps
3:05feel premium is illustrations. Now I
3:07want to be very upfront about something.
3:09Illustrations used to be my secret
3:10weapon to making my [music] apps stand
3:12out. And then ChatGPT image came out and
3:14now every single app has illustrations.
3:16So just having illustrations is not
3:18enough. You can check that box, but your
3:20app is still going to feel vibe coded.
3:21[music] So now when I build an app and I
3:23add illustrations, I ask myself, "How
3:24can I do this better? How can I make
3:26sure that my illustration stand out
3:28among other apps?" Let me show you what
3:30I mean. This is the onboarding for my
3:32calorie tracking app Amy. People keep
3:33messaging me about this onboarding
3:35saying this is one of the best
3:36onboarding that they've seen. I think
3:37someone actually signed up twice just to
3:39go through the onboarding. Now think
3:41about that for a second. Is there an
3:42onboarding flow you actually enjoyed
3:44going through? [music] Probably not. So
3:45for someone to compliment mine, I was
3:47clearly doing something right. And I
3:48think it all came down to these really
3:50great playful illustrations. Now here's
3:52the exact same onboarding if I took a
3:55one-shotted GPT image and just [music]
3:57replaced it in the onboarding. It's the
3:58same exact onboarding on paper, but it
4:01just feels different. [music] One feels
4:03premium and then one feels like every
4:04other onboarding that was shipped this
4:06year. The difference is my wife Cecilia
4:07actually hand drew these illustrations
4:09in Procreate. But more importantly, we
4:11spent a lot of time iterating and trying
4:14to find the illustration style. [music]
4:16And the bar we were holding ourselves to
4:17was we wanted someone to say when they
4:19saw these illustrations, "How did they
4:21possibly prompt that?" Or even better,
4:23questioning whether this was
4:24AI-generated at all. Now if you can't
4:25work with an artist and you're
4:26generating with AI, here's the secret to
4:28getting really good illustrations.
4:29[music] It's that one or two prompts is
4:31not enough. Because if you can get this
4:33result in one or two prompts, that means
4:35everybody else can, too. And that's why
4:37certain illustrations feel very generic
4:39because they've been overdone by like
4:40every single [music] app. In my
4:41experience, the best illustrations I've
4:43come up with have come after like
4:4530-plus prompts and a ton of iteration.
4:46[music]
4:47This involves taking multiple styles,
4:49mixing them together, matching them up,
4:50trying to get creative with it. One
4:52thing I like to do is look at random
4:53sites like this is a random Midjourney
4:55Sref reference thing. I won't explain
4:57what that is, but basically it has a
4:59bunch of different illustration [music]
5:01styles on this website, and I love going
5:03on here for inspiration. But again, you
5:04don't want to go here and just blindly
5:06copy and paste a style. What you want to
5:08do is just get inspiration. You want to
5:10mix and match styles. You want to try to
5:12figure out how to make it your own. I
5:13think a great example of a company that
5:15does this is Duolingo. Could ChatGPT
5:17generate an asset that looks like
5:19mascot? Yeah, it definitely could, but
5:21there's a lot of subtle intention behind
5:23the designs that even though it does
5:25look like some AI-generated assets, it
5:28does feel very unique. It does feel like
5:30Duolingo. It doesn't feel like this was
5:32made in one or two prompts, and that
5:33should be the target for you. [music] If
5:35you look at the onboarding, another
5:36thing that's going on here is that all
5:38of the illustrations are animated. Now
5:40in my case, Cecilia also hand animated
5:42these in Procreate. But if you can't do
5:44that, here is a hack that most people
5:46don't know about. You can actually use
5:47Midjourney to animate [music] images,
5:49and that's my favorite way to do
5:51animations. All you have to do is throw
5:52a static image into Midjourney, hit
5:55loop, and then tell it what kind of
5:56animation you want, and it's going to
5:58give [music] you four variants. And you
5:59can keep repeating this process and
6:01remixing until you get something you
6:03like. For example, I threw the Amy
6:04mascot in here, looped it, and now I
6:06have this really cool animation of Amy
6:08moving. But again, similar to the
6:09animations, the takeaway here with the
6:11illustrations is you have to put in
6:14[music] effort. You need prompting a
6:15lot, remixing, and trying to make your
6:17own style that stands out. Now real
6:19quick, because there is a theme here,
6:2030-plus prompts on the illustrations, 6
6:22hours working on some simple animations.
6:24Whatever you're polishing, you're going
6:26to be iterating a lot. And the best way
6:27I found to do this is to have a
6:29conversation with your agent. I dictate
6:31everything, and the tool I recommend for
6:32this is WhisperFlow, and a huge
6:34shout-out to them for sponsoring this
6:36video. WhisperFlow is smart
6:37voice-to-text that works in any app. So
6:39when I'm iterating on illustrations, I'm
6:41literally talking to ChatGPT and asking
6:43it,
6:44"Can you modify the image so it's a
6:45little bit more minimal? I don't like
6:47how the edges of the illustration look."
6:49Probably doesn't seem that annoying to
6:51type this out, but try doing that 30
6:53times in a row. It does get pretty old.
6:55Dictating is just so much better. Now
6:57another reason I recommend WhisperFlow
6:58is it's perfect for developers because
7:00it understands technical terminology. So
7:02when I say something like,
7:05"Use state, convex schema, webhook
7:07handler,"
7:09>> [music]
7:09>> it gets it right every time. And if
7:11you're using Cursor or Windsurf, it
7:12works with these IDEs. So it understands
7:15what's going on in the code you're
7:16looking at, and it can even tag files.
7:18Check this out, if I say,
7:20>> [music]
7:20>> "Please enhance the styles from
7:21subscription overview.tsx,"
7:25it'll actually tag the file without me
7:26touching the keyboard. It's available on
7:28desktop, so you can use it inside of
7:30Cursor, Claude Code, or ChatGPT like I
7:32do. And it's on iOS and Android as well.
7:34I'll leave a link in the description,
7:35and there's a code for 1 month free, and
7:37that's on top of the existing 14 days of
7:39WhisperFlow Pro you get when you sign
7:41up. They did not have to provide that.
7:43I'm the one that asked for that. So a
7:44huge thank you to them for doing that.
7:45Okay, now back to the illustrations
7:46because I do have one more
7:48recommendations. And unfortunately, it's
7:50something that a lot of you guys are not
7:52going to like hearing. If you really
7:53want to stand out and have an edge, it's
7:55to work with an actual artist. [music]
7:57Now before you say, "Okay, well that's
7:58easy for you to say. Your wife is a
8:00designer." Just know that even before
8:01working with Cecilia, I had already
8:03commissioned graphic designers in the
8:04past. And I will still continue doing
8:06that, especially if there's a specific
8:08art style that I don't think she's
8:10suited for. It's actually way more
8:11affordable than you think. And I've
8:12gotten really good graphics commissioned
8:14for like a few hundred dollars. Now
8:16there are two reasons why I think it's
8:18worth it. The first is that your odds of
8:19getting something truly unique are
8:21exponentially higher when you work with
8:23a human rather than when you just prompt
8:25it. Now you can absolutely escape the
8:27default AI generated look if you use
8:29enough prompts. Or you can work with an
8:30artist and know for a [music] fact that
8:32it's original. Now the second reason,
8:33and it's something that I think people
8:35massively underestimate, is that the
8:37story itself of how you build your app
8:39also contributes to making it feel
8:41premium. The moment people find out that
8:43your illustrations were hand-drawn by an
8:45artist, the entire app instantly feels
8:47different. Even if AI could replicate
8:49the asset. I mean, think about watches.
8:5190% of people cannot tell the difference
8:53between a hundred-dollar watch and a
8:54thirty-thousand-dollar watch. But the
8:56people who can tell the difference care
8:57about one thing, and it's craftsmanship.
9:00Somebody spent hundreds of hours
9:01handcrafting that thing. At that point,
9:03it no longer becomes a watch. It really
9:05becomes a story or an experience. And I
9:07genuinely think this is where software
9:09is heading in a world where AI can just
9:10generate a million apps. Knowing the
9:12story behind the app and that a human is
9:15the one that actually crafted some of
9:17the assets is actually going to start
9:18mattering to people. If you're on a
9:19budget, here is a hybrid workflow. What
9:21I would do is commission an artist to
9:23make a base style or mascot. And again,
9:26you could get this done for like a few
9:28hundred dollars. And then use [music] AI
9:29to create derivatives or variants from
9:31it. This is actually what I did for my
9:32unreleased app Lily. I had Cecilia
9:34hand-draw the asset of the mascot. And
9:37then I created a bunch of derivatives
9:39using chat GPT. [music] You get the best
9:40of both worlds where you get an original
9:42commissioned piece from an artist where
9:43there's a really good story, but you
9:44also have an affordable way to just make
9:47a bunch of variants. Okay, so far we've
9:48covered animations and illustrations,
9:50but craft is not just UI. [music] Craft
9:52and taste goes way beyond visual, so let
9:54me show you two examples of what I mean
9:56by that. My calorie tracking app Amy has
9:58a feature where you can take photos of
9:59food, and then it'll calculate the
10:01calories [music] from that. In the first
10:02version I shipped, you tap a button, and
10:04then the default iOS camera is going to
10:06appear for you to take [music] the
10:07photo, which is totally fine. It was
10:08like that for months. Nobody had a
10:10problem with it. One thing I'm always
10:11doing is looking for inspiration and
10:13studying other apps, and then one day I
10:15came across this [music] in my timeline.
10:17It's that chat GPT shipped a new inline
10:19camera feature. So, instead of using the
10:21default iOS camera, when you click the
10:22camera button in their app, it's going
10:24to have this inline camera experience
10:26pop up. It is just such a good
10:28experience. The second I saw this, I was
10:30like, I have to try to recreate this and
10:32see if I can bring this into Amy. So, I
10:34spent a few hours reverse engineering
10:35the animation, trying to figure out why
10:37does it feel so good and smooth, and how
10:38[music] do I keep it performant, and I
10:39ended up actually shipping it. Now, 99%
10:42of users are not going to notice this
10:43change at all. But, after you've used it
10:44like five or 10 times, you kind of start
10:46[music] noticing when it's absent in
10:48other apps. So, if they use any of my
10:50competitors and they click the camera
10:51button, and then they just see the
10:52default iOS camera appear, they're
10:54probably going to think to themselves,
10:55why does this just feel a little bit
10:57different than Amy's? Unless they're
10:58really perceptive, they probably can't
11:00name exactly what the difference is, but
11:02I guarantee that [music] they can feel
11:04it. Also, I want to point out, as you
11:05probably noticed, I did not ship this on
11:07day one. Version one was the default iOS
11:09camera experience, and then I came back
11:11months later and then tried elevating
11:12the experience. [music] And that's
11:13actually a tip on its own. You don't
11:15have to build a premium version in
11:16version one. You can layer this over
11:19time. Now, the second example of craft I
11:20want to show you guys is an invisible
11:22one. It's where there's no UI at all.
11:24Amy has a dictation feature, so you can
11:26click a button and then just start
11:27dictating what you ate, and then it's
11:29just going to apply it as an entry. And
11:31it works great. For 90% of the entries,
11:33it's just going to get it. But I did
11:34notice there was an edge case when
11:36there's a restaurant that's really
11:38[music] tough to say. There's a Korean
11:39restaurant in Dallas called Saemaul, and
11:41my app was not getting it. So, I made a
11:43quick change to hopefully improve this
11:44where when you use dictation, Amy is in
11:47the background going to get the 30
11:49closest restaurants and add it as
11:51context. So, it has a higher shot of
11:53getting the name if you dictate it. So,
11:55if I'm at the restaurant and I dictate
11:56and say Korean barbecue from Saemaul, it
11:58actually is going to get it right. It
12:00seems like such a small deal, but it's
12:01going to be one of those things where
12:02when the app gets it, the user is 100%
12:05going to say, "Wow, I can't believe it
12:07got that." I absolutely love adding
12:09features or enhancements like this that
12:10pleasantly surprise users. That's
12:12another way to stand out and feel
12:14premium. Also just wanted to say that
12:15this location-based feature is
12:17completely opt-in. You have to turn it
12:18on in the settings just before anyone
12:20calls me out on that. But let's zoom out
12:21because you probably notice a pattern by
12:23this point. Every single example in this
12:25video started the exact same way. It was
12:27with something already working, and then
12:29I just asked myself, "How can I make
12:31this better? How can I elevate this?"
12:33And to be honest, that's the biggest
12:34thing that I'm doing when building apps,
12:35and I think that's the secret. Is I'm
12:37constantly trying to push myself. And I
12:39think one of the biggest positive
12:40benefits of AI coding is that a lot of
12:42the basics are basically taken care of.
12:44So, now I and you can focus on all of
12:47the small details. That's where we can
12:48be pouring our energy into. When you
12:50spend 6 hours on an animation or 30-plus
12:53prompts on an illustration or hours
12:55studying trying to reverse-engineer a
12:57very simple-looking camera animation,
12:59you do it enough times, you start seeing
13:00patterns, you get better at
13:01experimenting, you start developing
13:03better techniques, [music] you start
13:05building taste. And a lot of people way
13:07smarter than me are saying that taste is
13:08going to be the defining skill for the
13:10next era, and I really believe that's
13:12true. If you're going to take anything
13:13from this video, here are three things.
13:15Number one is to always ask yourself,
13:17"Can I make this better?" It's important
13:19to remember that AI's first output is
13:21the starting line. It is not the finish
13:23line. Number two, it's that unique and
13:25good take iteration. Whether that's
13:27iterating with a 100 plus prompts or
13:29working with a human artist. And number
13:31three, it's that premium is felt more
13:33than seen. [music] Some of the best
13:35craft is completely invisible or just
13:37will go unnoticed by 99% of [music]
13:39users. I'll leave a link in the
13:40description to a couple more videos
13:42where I talk about craft including a
13:43conversation with Greg Eisenberg where I
13:45talk about some of the techniques like
13:47fonts and illustrations. And also if you
13:49like this content, check out my
13:50Instagram and TikTok. I post almost
13:52every other day about building
13:53productivity apps. I drop a lot of short
13:55form content on how to build premium app
13:57experiences. And obviously if you like
13:59it, don't forget to subscribe. But thank
14:00you guys so much for watching and I will
14:02see you guys in the next video.
14:06>> [music]

About this transcript

This page contains the full transcript of How I Make Apps FEEL Premium (5 examples) by Chris Raroque, generated from the public captions YouTube serves with the video. The transcript has 3,371 words across 485 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.