0:02 Another day, another startup. Recently,
0:05 I moved to Malaysia and from there, I've
0:06 pretty much been spending most of my
0:08 time coding my latest project. And it's
0:10 been super fun hacking away at something
0:12 new again. And luckily, I've been
0:14 documenting the whole journey from
0:15 coming up with the app idea to building
0:17 the app to getting that first paying
0:19 customer. In this video, I'll show you
0:21 every step of the process so you get an
0:23 idea of what it takes to build and ship
0:26 a profitable app from scratch completely
0:27 solo. And feel free to stick around
0:29 until the end to see how much money my
0:32 app actually made. My app idea is just a
0:34 simple tool that allows me to download
0:36 video transcripts in bulk from YouTube
0:37 channels or playlists. And you're
0:39 probably asking, why do I want a tool
0:41 like this? Well, it's no secret that
0:43 YouTube is a gold mine of knowledge and
0:45 information, but it's all trapped in
0:48 video format. So, I'm building this tool
0:50 for one main reason. I want a quick and
0:52 easy way to extract all this gold from
0:55 YouTube videos in text format so that I
0:57 can feed this data directly into my own
0:59 custom AI models to help me with
1:01 everyday tasks like writing better
1:04 YouTube scripts or creating custom AI
1:06 agents to help me with my startups. Then
1:08 after giving it some more thought, I
1:09 know I'm not the only one who needs this
1:11 tool. Pretty much anyone wanting to
1:13 train their own custom AI models with
1:15 YouTube content is facing the same
1:17 problem. Manual transcript extraction
1:19 from YouTube is painfully slow and
1:21 tedious and currently there aren't many
1:22 good solutions out there. So then I
1:24 thought why not try building this tool
1:26 myself. It solves a real problem I am
1:28 facing and it could potentially be
1:29 another stream of income if I execute
1:31 and market it well to the other users
1:33 who are facing the same problem. But
1:35 before I start building, I need to do
1:36 some quick market research to see what
1:39 my competitors are offering. First,
1:41 there's YouTube transcript.io IO, which
1:43 offers bulk downloading but requires a
1:45 subscription with a credit system. So, I
1:47 can definitely see an opportunity for my
1:49 app to compete by offering a simpler
1:50 pricing model. And from looking at their
1:52 landing page, they report having over
1:55 350,000 users worldwide, which is a
1:57 clear sign there's already real demand
1:59 for this type of product. This gives me
2:01 even more confidence in my idea and
2:03 extra motivation to compete with them to
2:05 take a slice of the market. There are
2:07 also other sites like Note GBT and
2:09 Tactic, but those sites only let you
2:11 download transcripts. one video at a
2:13 time. So, my approach will be simple.
2:14 Create a tool with a simpler pricing
2:16 model and a super clean interface that
2:18 makes downloading YouTube transcripts in
2:21 bulk cheap, quick, and painless. After
2:23 scouting out my competitors, I was ready
2:25 to jump into my code editor and start
2:27 building. But before I could write a
2:29 single line of code, I had to plan out
2:30 what the first version of my app would
2:33 be like. Otherwise, I'll either end up
2:34 building something that's not useful, or
2:36 I'll constantly get distracted adding
2:38 shiny new features that don't actually
2:40 solve the root problem. I prefer using
2:42 the SLC framework to plan the early
2:44 versions of my app. I'm not a fan of
2:46 shipping MVPs that are buggy at launch
2:49 and don't do the key one to two features
2:51 they promise just for the sake of speed
2:53 or shipping fast. The SLC framework
2:55 stands for simple, lovable, and
2:56 complete. And I think it's a much better
2:58 approach than just building something
3:00 that's minimum and viable. For my app to
3:02 be simple, it needs to have one
3:04 straightforward path to solving the root
3:06 problem, and that's extracting YouTube
3:09 transcripts in bulk. This means users
3:10 should be able to complete the task in
3:13 as few steps or clicks as possible.
3:15 Nobody likes using a halfbaked product
3:17 for version one. This means users should
3:19 be able to download thousands of
3:21 transcripts quickly. My app won't be
3:23 feature- packed at launch, but it will
3:25 be complete. This means users should be
3:27 able to sign up, make a purchase, and
3:29 immediately download transcripts. This
3:31 is the leanest possible version of my
3:33 app and the goal is to provide value
3:35 from day one and then iterate the
3:37 product based on real feedback from
3:39 users. Like I've said before, I'm aiming
3:41 to ship a version one of something
3:43 simple instead of a version 0.1 of
3:45 something complex. Now that I have a
3:47 basic plan in place, it's time to commit
3:49 and secure a domain for the app. I
3:50 decided to go with YouTube
3:52 transcripts.io as it's short,
3:55 descriptive, SEOfriendly, and doesn't
3:57 break the bank. Now that the domain was
3:58 purchased, it was time to start
4:00 building. For this project, I'm using
4:03 the standard web text stack, React and
4:05 Next.js for my front end and API,
4:07 Superbase for my database and
4:09 authentication, and Stripe to handle
4:11 payments. Normally, setting up my
4:13 projects would take me a few hours to
4:15 get everything connected and working.
4:17 But this time, I am using Tempo to help
4:19 me with the setup process, and I was
4:21 able to get my OR, database, and
4:23 payments all working in under 10
4:24 minutes. All I had to do was click
4:26 through their guided setup, paste in a
4:29 few API keys and within minutes I had a
4:31 fully functioning project with all the
4:33 annoying stuff correctly configured.
4:35 Then I gave a simple prompt to Tempo
4:38 describing what my app does. And then it
4:39 just starts generating the project
4:41 files, documenting project requirements,
4:44 and building out the UI. And honestly,
4:46 the first version of the landing page
4:48 tempo generated was pretty impressive.
4:50 It wasn't perfect, but it was a solid
4:51 starting point with a clean layout and
4:54 some decent copy. From there, I gave the
4:56 AI a few more prompts to refine some
4:58 sections, like updating the feature
5:00 section to better highlight what the app
5:02 actually does, adding a use cases
5:04 section and an FAQ section to answer
5:06 common questions or objections up front.
5:08 Before we move on, let me give you a
5:10 quick tour of the Tempo UI. The
5:12 interface is split into three tabs:
5:14 product, design, and code. In the
5:16 product tab, you can write out your
5:18 product requirements, features, and even
5:20 create full user flow diagrams of your
5:22 app. The most useful thing about this is
5:24 that all this information gets fed
5:27 directly to the AI as context, which
5:29 reduces hallucinations massively. In the
5:31 design tab, you can build out your UI
5:33 just by clicking and dragging elements
5:35 pretty much just like any other design
5:37 tool like Figma or Web Flow, but it
5:39 generates actual React code behind the
5:41 scenes. They also support importing your
5:43 custom Figma designs, so you can use
5:45 them as a starting point and ask the AI
5:47 to generate React code for you based on
5:49 the designs. Finally, the code tab is
5:51 where you can open and edit any file in
5:53 your project just like you would in VS
5:55 Code or cursor. It also integrates
5:57 nicely with GitHub, so you can push your
5:59 code directly to a repo and keep
6:01 everything version controlled. With the
6:03 initial setup all done, I push the code
6:05 to GitHub and fired up Cursor so that I
6:07 can start building out the core features
6:09 of my app. So, enjoy this time-lapse of
6:11 me hacking away over the next couple
6:13 days whilst I get this app ready for launch.
6:35 [Music]
7:10 [Music] Heat. Heat.
7:44 After days of non-stop coding and
7:46 testing, I think the app is finally
7:48 ready. It's time to launch and get some
7:50 users. My plan for the initial launch is
7:52 pretty basic. I'll post on all of the
7:54 typical launch platforms like Product
7:56 Hunt, Hacker News, and Twitter to try my
7:59 luck going viral, but I'm not banking on
8:01 this too much. My tool is quite niche
8:03 and solves a very specific problem. So
8:05 there's really no point trying to appeal
8:07 to a wider audience at this stage. And
8:09 hot take, I don't think Product Hunt is
8:10 actually a good platform to launch for
8:12 most indie hackers. Unless you're making
8:14 a product specifically targeted to other
8:16 founders because the truth is regular
8:18 customers aren't going to Product Hunt
8:20 to find tools to solve their problems.
8:22 It's usually just a bunch of other
8:23 founders looking for a place to share
8:25 their products. But you do get a nice
8:27 backlink for SEO and a bump in traffic
8:29 if you can rank highly. So, it's at
8:31 least worth trying. That's why I'll be
8:33 spending most of my time going directly
8:35 to my ideal customers and building
8:36 relationships with them first. This
8:38 means finding niche subreddits and
8:40 forums and sharing my launch there
8:42 instead. So, how did the launch go and
8:45 how much money has my app actually made?
8:47 Well, I can tell you my app didn't blow
8:48 up overnight. And I'm not going to lie,
8:50 it was discouraging at first seeing my
8:52 launch get less traction than I hoped.
8:54 It's disappointing, but this won't be
8:55 one of those videos where I launch my
8:57 product and it magically goes viral. And
8:59 honestly, I think sharing realistic
9:01 launches like these can be more valuable
9:03 than the overnight success stories
9:05 because whilst they can be inspiring,
9:07 they can also create unrealistic
9:09 expectations and discourage you to keep
9:11 going if you don't see immediate
9:13 results. So, even though the launch
9:14 wasn't what I hoped for, I'm not going
9:16 to let it distract me from improving and
9:18 growing my product until I can get it to
9:20 a point where I'm satisfied and feel
9:21 like there's nothing more I can do.
9:23 Since the launch, I've managed to get 13
9:26 user signups with six of them converting
9:28 to actual paying customers. And so far,
9:30 my app has earned over £200 since the
9:32 launch. It's not some life-changing
9:34 money, but at least it's proof that
9:35 people are willing to pay for my
9:37 product, even at its earliest stage. So,
9:40 what's next? I have a bunch of ideas and
9:41 improvements I want to make to this app
9:42 already. And of course, I'll be
9:44 documenting everything on this channel.
9:46 So, if you want to see how I grow this
9:48 app from these humble beginnings into
9:50 something bigger, feel free to subscribe
Chrome Extension