My Favourite AI Workflow Nobody Talks About — Transcript
Full transcript
- 0:00This is my favorite AI workflow right
- 0:01now. It is so powerful, and you don't
- 0:03even need to know how to code, although
- 0:04it does help. And it is essentially
- 0:05free, or at least much, much cheaper
- 0:07than paying for most software
- 0:09subscriptions. Let me explain by just
- 0:10showing you guys first some use cases,
- 0:12and then I will break it down
- 0:13step-by-step with a five-step framework
- 0:15on how to do it this AI workflow. Sounds
- 0:17good? A portion of this video is
- 0:19sponsored by Bolt.
- 0:22Use case number one, which was the bane
- 0:24of my existence, accounting. If you
- 0:26yourself are a small business owner, or
- 0:28know somebody that is a small business
- 0:30owner, you may understand what I mean
- 0:31when I say that accounting is the bane
- 0:33of my existence. Stuff like bookkeeping,
- 0:35filing taxes, payroll, filling out a
- 0:37million forms about a million things. It
- 0:39causes me physical pain just thinking
- 0:41about it.
- 0:43Especially because I do a lot of
- 0:45cross-border work, meaning I hire people
- 0:47from all over the world, and we also
- 0:48have clients from all over the world.
- 0:50So, I have this very specific kind of
- 0:51setup that is not optimal for most
- 0:53commercial accounting software. Except
- 0:55now, because we actually built our own
- 0:57hyper-specific AI app to do accounting
- 1:00for us. It is so custom to this company.
- 1:02It's able to deal with specific
- 1:03regulations from specific countries. It
- 1:05can interpret different kinds of
- 1:06currencies, and is able to do
- 1:08bookkeeping and forms and things like
- 1:09that that abide by the laws of multiple
- 1:12different countries as well. We're also
- 1:13using open-source AI and hosted locally,
- 1:15so it's not leaking things to the
- 1:17internet, and it's free, no matter how
- 1:19many documents, receipts, and invoices,
- 1:21and whatever that I want to feed through
- 1:22it. Commercial software, by the way,
- 1:23usually costs around like $15 to $30 per
- 1:25month. Use case number two, making
- 1:27custom slide decks. If you've been to
- 1:29any of my live streams, or have done any
- 1:31of our courses or workshops, you would
- 1:32have seen this very specific style of
- 1:34slides that we use. While previously it
- 1:36would have taken ages to make slide
- 1:38decks for every single presentation, now
- 1:40we built this super custom slide deck
- 1:42maker where we can feed in the
- 1:43information and have it just come out
- 1:45with these very beautiful slides, so
- 1:47streamlined and custom to exactly what
- 1:48we want, and also free. Use case number
- 1:51three, another example. Couple months
- 1:53back, for whatever reason, I I that I
- 1:55wanted to make a manga. But the problem
- 1:57is that I don't know how to draw. Like
- 1:59this is my drawing level. Seeing as I
- 2:01did not have the patience,
- 2:02unfortunately, to spend months and
- 2:03months drawing, which by the way,
- 2:05totally cool if that is what you want to
- 2:07do. But for me, it was kind of like a
- 2:08bucket list thing where I just wanted to
- 2:10make a manga. So, I made, you guessed
- 2:12it, a hyper-specific custom AI app that
- 2:14was able to make an entire manga just
- 2:16based upon text prompts and my terrible
- 2:19drawings. I have so many more examples.
- 2:21Let's see, what else have I made? Here's
- 2:23another custom email AI agent that's
- 2:24able to screen my emails and pre-draft
- 2:26them for me. And about a year ago, I
- 2:28also made another one that is able to
- 2:29handle like my seven calendars and tell
- 2:31me what I'm supposed to do every day.
- 2:32Yes, I have so many other examples, but
- 2:35I think you get my point. So, these are
- 2:37examples of my favorite AI workflow
- 2:39right now, which I don't really know
- 2:41what the name is for this. So, I shall
- 2:43call it hyper-specific apps. Definition
- 2:46of hyper-specific app is when you use
- 2:48AI-assisted coding to build software for
- 2:51very specific personal or business use
- 2:53cases that would never exist as
- 2:55commercial software. Here are the pros.
- 2:57I think the biggest pro of building a
- 2:59hyper-specific app is that you're able
- 3:00to build custom features that are so
- 3:03specific to your use case that it's
- 3:04impossible for you to actually find from
- 3:06commercial software. Another very
- 3:08attractive pro is that it is essentially
- 3:10free, minus the building and setup
- 3:12costs, especially if you use open-source
- 3:13models. Like on average, these apps cost
- 3:15me around $10 to $20 to build. And
- 3:17depending on where I'm hosting these
- 3:18apps, like where these apps live, you
- 3:20can run them again and again without
- 3:22incurring additional costs, which I will
- 3:24actually talk about like hosting and
- 3:25stuff like that a little bit later in
- 3:26the video. But yes, for now, just
- 3:28realize that at the very minimum, it's
- 3:29going to be a lot cheaper than buying
- 3:31like a monthly subscription for most
- 3:32software. With building your own apps,
- 3:34you also do have full control over your
- 3:36software. So, you don't need to be
- 3:37worried about like where it's sending
- 3:38its data, what's happening to it, things
- 3:40like that. And finally, you don't need
- 3:42to write code yourself to build
- 3:44something like this. Don't get me wrong,
- 3:45I think you're interested in learning
- 3:46how to code, you absolutely should still
- 3:48do it because there is a ceiling to what
- 3:50you can actually build without knowing
- 3:51how to code, and overall, it would also
- 3:53save you a lot more money if you knew
- 3:54how to code. But you don't need to know
- 3:55how to code when you're starting out
- 3:57building your hyper-specific apps, which
- 3:58is pretty cool. I'll explain the process
- 4:00in a bit. But first, I do want to
- 4:01mention some cons as well, some caveats.
- 4:04The first one is that it does take time.
- 4:06On average, it takes me about 2 to 3
- 4:07hours up to a day to build the first
- 4:09version of these hyper-specific apps.
- 4:12But there is a little bit of a learning
- 4:13curve. So when I was just starting out,
- 4:15or if you're someone who has no idea how
- 4:16to build anything, zero engineering
- 4:17background, no coding skills, it might
- 4:19take you a little bit longer, like maybe
- 4:20a few days to a week to get started
- 4:22properly. So some time and effort, yes.
- 4:24Also, it could require a little bit of
- 4:25setup and building cost. Specifically,
- 4:28if you are going to be using AI-assisted
- 4:29coding, you're probably going to need to
- 4:30pay for a subscription to use a coding
- 4:33agent of your choosing. And if you want
- 4:34your app to be able to run continuously,
- 4:36you might have to pay for some type of
- 4:38cloud service or a local machine that
- 4:41will continuously run. Although, if
- 4:42you're okay using like open-source
- 4:44coding models that are not like as
- 4:45optimized for coding, um or you're okay
- 4:48with things like maybe not running all
- 4:49the time, you can technically also do
- 4:51this for free. All right, great. Let me
- 4:52now explain the five-step process for
- 4:55how to do this AI workflow to build a
- 4:57hyper-specific app. Step number one is
- 5:00to identify which workflow you want to
- 5:02build a hyper-specific app for. There's
- 5:05like three categories of things that my
- 5:06hyper-specific apps usually fall under.
- 5:08So hopefully, this is also helpful for
- 5:10you to come up with what you want to
- 5:11build. The first one are things that are
- 5:12the bane of your existence, but
- 5:14unfortunately, to exist, you must
- 5:17actually do it. Biggest example for me
- 5:19was accounting. I will put on screen now
- 5:21some other examples in this category.
- 5:23The second is things that you've been
- 5:24procrastinating on because it's like you
- 5:26know it's important, but you just don't
- 5:29want to do it cuz it's a lot of work and
- 5:31you just don't feel like doing it. But
- 5:32you know, if you did it, it would
- 5:33actually be much better for you. An
- 5:35example for me is tracking my exercise,
- 5:37my food, and my sleep. Things that I
- 5:39know that I should do, but it's just
- 5:41like effort, you know? I will put on
- 5:43screen some other examples in this
- 5:44category. And the third category is
- 5:46something that you want to do, but you
- 5:48can't or haven't because you lack the
- 5:50skill or you don't have the time and you
- 5:52just like maybe don't have the resources
- 5:53or energy to to dedicate to doing it.
- 5:56This could be like a fun hobby project.
- 5:57For example, for me it was making a
- 5:59manga, except it was impossible because
- 6:01I'm so bad at drawing. And I know for
- 6:02others like many people want to build
- 6:04like a software or a game, but they just
- 6:06like don't have the skills and time to
- 6:08do that. I'm going to put on screen out
- 6:09some other examples in this category.
- 6:11And I do think this is like a really
- 6:12interesting category because it
- 6:14literally enables you to do something
- 6:16that you simply could not do before.
- 6:21>> [screaming]
- 6:23>> That's pretty cool. All right, I hope
- 6:24this helps you decide what you may want
- 6:26to build a hyper-specific app on. And do
- 6:28put in the comments what you want to
- 6:29build. I genuinely really want to know
- 6:31what you guys are interested in
- 6:32building. Step number two is to map out
- 6:35the process workflow. As in like map out
- 6:37the current existing workflow. This step
- 6:40can take a lot of time or very little
- 6:41time depending on how well you
- 6:43understand the process. Let me explain.
- 6:45For example, for the accounting stuff
- 6:46that I wanted to build this
- 6:47hyper-specific app for, a core feature
- 6:49would be bookkeeping. And this is
- 6:50something that I already understand the
- 6:52full workflow of because I've been doing
- 6:53it for like 5 years. Every month I
- 6:55needed to one, collect all of the
- 6:57receipts and invoices. Two, match them
- 6:59to all of my corporate credit cards and
- 7:01also my bank statements. And then three,
- 7:03have a final deliverable as a
- 7:04spreadsheet with all of the entries as
- 7:06transactions and the supporting
- 7:08documents for each transaction. And that
- 7:10would be the mapped out process. Didn't
- 7:11take me a lot of time to figure this out
- 7:12because I had already done it so many
- 7:14times. But for mapping out the workflow
- 7:15for generating for example, I literally
- 7:18had no idea the process of making manga.
- 7:20So in order to map out this workflow, I
- 7:22actually have to go take an entire
- 7:23course from an expert to map out the
- 7:25process of what it takes to make a
- 7:27manga, which in case you're curious is
- 7:29this. So after you map out your
- 7:30workflow, we can then move on to step
- 7:32three, which is identifying where your
- 7:34hyper-specific app would fit in in that
- 7:36workflow. As amazing as technology and
- 7:38software and AI is these days, it's
- 7:40still not possible to just like take
- 7:42your entire workflow that you just
- 7:43mapped out and stick it into AI and just
- 7:45tell it to like build the entire thing,
- 7:46unfortunately. And also, maybe you don't
- 7:48actually want that, right? Like part of
- 7:50the process, you might want it to remain
- 7:51manual. Or there's no need for you to do
- 7:53that. So, that's why we do want to
- 7:54explicitly go in there and identify
- 7:56which parts of the workflow we want to
- 7:57be part of the hyper-specific app. Let's
- 7:59use the manga example here. So, here's
- 8:00the workflow for making manga, which I
- 8:02learned from taking that course from a
- 8:04professional mangaka. I learned that
- 8:05there were nine steps to making a manga.
- 8:07One was to pick a topic, including
- 8:08central values and central themes that I
- 8:10wanted to address. Two was to design the
- 8:12characters. Three was to have a major
- 8:14plot synopsis, like an explanation for
- 8:16what the manga should be about. Four is
- 8:18to write out the entire story and split
- 8:20them into chapters and scenes. Five is
- 8:22then to transform these scenes and
- 8:23chapters into panel descriptions, like
- 8:26what each little panel drawing should be
- 8:28portraying. Six is to actually sketch
- 8:30out these scenes and draw them and then
- 8:32design the panels as well, like the how
- 8:34the panels are all supposed to fit
- 8:35together and how the pages are exposed
- 8:37to be divided. I was really surprised by
- 8:38how much I thought and detail and
- 8:40process went into just constructing how
- 8:42the panels fit together. And seven is to
- 8:44fill in the details of these sketches,
- 8:45like actually draw them nicely. Eight is
- 8:47to then finally bind them all together
- 8:49into one whole manga entity. And nine is
- 8:52to go publish it. So, that is the entire
- 8:54workflow, but I don't actually want to
- 8:55include this entire workflow in the
- 8:58hyper-specific app itself. Because steps
- 9:00one to five actually involve a lot of
- 9:02writing and I actually really like
- 9:03writing and I think it's really fun.
- 9:05It's also where a lot of creativity
- 9:06comes in. So, I don't actually want this
- 9:08to be like auto-generated for me. I
- 9:10might like use AI to help me, you know,
- 9:12flesh out some stuff or like bounce
- 9:14ideas and stuff like that, but I do want
- 9:15to keep this process manual. Now, I do
- 9:17need a lot of help with and what I do
- 9:19want this hyper-specific app to be able
- 9:21to do is steps six, seven, and eight.
- 9:24Cuz that's like the visual parts, like
- 9:25sketching out the scenes, designing the
- 9:27panels, filling the details, and then
- 9:29binding them all together. This is like
- 9:30the blocker for me. It is impossible for
- 9:32me to do this because I don't know how
- 9:33to draw. So, that's why I want my
- 9:34hyper-specific app to be able to take
- 9:36care of these steps for me. And finally,
- 9:38step number nine, which is the publish,
- 9:40I don't actually care that much about
- 9:41that cuz I'm just doing it for my own
- 9:43like bucket list enjoyment. Amazing,
- 9:45great. You now know what you want your
- 9:47hyper-specific app to do. We can now
- 9:49move on to the next step, which is
- 9:50actually building your hyper-specific
- 9:52app using good AI coding practices. I've
- 9:55actually made an entire video on how to
- 9:57use AI coding in order to build apps
- 9:58both using code and no-code tools. So, I
- 10:01do recommend checking that out. I'll
- 10:02link it over here. I also made a videos
- 10:03where I show you guys me building these
- 10:05apps and I'll link them in the
- 10:05description as well. So, do check those
- 10:07out when you get serious on building
- 10:08your app, but I do want to give you a
- 10:09very fast quick start guide in this
- 10:11portion of the video, too. So,
- 10:12basically, three steps. You want to take
- 10:14what you want your hyper-specific app to
- 10:15do, and then change that into what is
- 10:17called a product requirement document.
- 10:19This is a standard document for defining
- 10:22the purpose, features, functionalities,
- 10:23and the behaviors of a product. This is
- 10:25what you're going to be feeding into
- 10:26your AI coding tool, so it knows what
- 10:29you're supposed to be building. I'm
- 10:29going to actually link a prompt in the
- 10:31description, which you can just copy
- 10:32paste into like any chatbot of your
- 10:34choosing, whether that be like ChatGPT,
- 10:36Claude, Grok, Gemini, whatever you want.
- 10:38And it would actually like ask you
- 10:39questions to help you come up with this
- 10:42PRD. See, I got you. All right, so after
- 10:44you have your PRD, you then want to pick
- 10:46your AI coding tool to build your
- 10:47hyper-specific app. In the AI coding
- 10:49fundamentals video, I literally like
- 10:51have an entire section dedicated to
- 10:53explaining tools, so please go to that
- 10:55video for a lot more details. But
- 10:56basically, there are range of different
- 10:58AI coding tools that range from
- 11:00completely no-code to for pro
- 11:02developers, and those are good for
- 11:03specific types of use cases as well. For
- 11:05me personally, because this is literally
- 11:07my job, I have tested pretty much the
- 11:10full spectrum, like all of the different
- 11:12AI coding tools out there. And I would
- 11:13use different AI coding tools for
- 11:15different purposes. Say for example, if
- 11:16I know that this hyper-specific app I'm
- 11:18trying to build is going to be like a
- 11:19big, complex project with lots of
- 11:21technical details, maybe even like
- 11:22pretty obscure technical things included
- 11:24in it, and at some point I might even
- 11:26want to bring in other developers to
- 11:27help me with it, then I would go
- 11:29directly with a code-based pro developer
- 11:31tool. Like the accounting software, for
- 11:32example, I primarily use Cloud Code for
- 11:34that. And when the project became bigger
- 11:36and I wanted to do more parallelization,
- 11:38like having multiple coding agents
- 11:40working at the same time and also like
- 11:41working with other developers, I
- 11:42switched over to Warp, which is really
- 11:44good for that kind of stuff. But, for
- 11:45the hyper-specific apps I'm making
- 11:47slides and also for the generator,
- 11:49these are web-based well-scoped projects
- 11:51that have pretty standard integration
- 11:53needs. So, I actually built these using
- 11:54Bolt, who I'm actually really excited to
- 11:56say is the sponsor of this portion of
- 11:57the video. I really like their product
- 11:59as a web-based option. using them for
- 12:01over a year and a half now. So, thank
- 12:04you so much for reaching out to sponsor
- 12:05this portion of the video, Bolt. Bolt is
- 12:07my go-to for creating really fast
- 12:08AI-powered software and fully functional
- 12:10web apps. I built this generator
- 12:12and the slide generator using Bolt. It
- 12:14lets you go from an idea to a fully
- 12:16deployed production-ready app just by
- 12:17describing what you want in plain
- 12:19English. It's crazy cuz it's not just
- 12:21the prototype. It's like an actual
- 12:22working app with a real back-end and
- 12:24running entirely on your browser. They
- 12:26also dropped two new features that I
- 12:27think will be really relevant for you if
- 12:29you want to build hyper-specific apps.
- 12:30The first one is image generation right
- 12:33inside Bolt. So, if you need a
- 12:34thumbnail, a mock-up, or any visual
- 12:36assets for whatever it is that you're
- 12:37building, you don't have to jump between
- 12:38tools anymore. You can just generate it
- 12:40directly within the platform itself. The
- 12:41second is connectors, and this is what
- 12:44I'm most excited about. Bolt now
- 12:45directly plugs into platforms like
- 12:47GitHub, Notion, Linear, Jira, and many
- 12:49more. So, your app is able to pull from
- 12:50real data, not just placeholder content.
- 12:53Now, that is what it takes from a
- 12:54prototype to something that you actually
- 12:56can ship. Over 7 million people are
- 12:58using Bolt now, including teams from
- 13:00Fortune 500 companies. It is free to
- 13:02start and no credit card required. Link
- 13:04is in the description. Thank you so
- 13:05much, Bolt, for sponsoring this portion
- 13:06of the video. Now, back to the video.
- 13:08Finally, to round out my brief guide to
- 13:09AI coding tools, I want to say that
- 13:11there are free options as well if cost
- 13:13is a really limiting factor, too. You
- 13:15can use open-source coding models. Just
- 13:17know that the coding experience itself
- 13:18is not going to be as optimized. And you
- 13:20need to be more careful about things
- 13:22like version control and and privacy and
- 13:24security things. So, I would only
- 13:25recommend this option to people who do
- 13:27have a little bit of coding experience
- 13:29already and a lot more patience. I'll
- 13:31put on screen some tools and models to
- 13:32check out as well here. Okay, great. So,
- 13:34now you've picked your AI coding tool.
- 13:35You want to take your PRD, remember your
- 13:37PRD from earlier, and stick that into
- 13:39your AI coding tool to get started on
- 13:42the process of building your
- 13:43hyper-specific app. Sorry for sounding
- 13:45like a broken record here, but in the AI
- 13:47coding fundamentals video, I go into a
- 13:49lot more detail about how to go through
- 13:51this building process. So, please do
- 13:53check out that video. The AI coding
- 13:54principles that I presented in that
- 13:56video, I have dubbed the tiny ferrets
- 13:58carry dangerous code framework, which
- 14:01stands for thinking, as in always be
- 14:03thinking and rethinking deeply about the
- 14:05features that you're building. You must
- 14:06not throw your PRD at your AI coding
- 14:08tool and just run away. Second is the
- 14:10focus on existing frameworks. You don't
- 14:12need to reinvent the wheel as you're
- 14:13building your app. There are most likely
- 14:16existing frameworks that other
- 14:17developers have already built. So,
- 14:19instead of trying to do something from
- 14:20scratch, help your AI agent provide
- 14:22relevant documentation and frameworks to
- 14:25your AI tool, so we can draw on the
- 14:26knowledge of past work. C stands for
- 14:28checkpoints. Create checkpoints as
- 14:30you're building your app, thinking about
- 14:31version control using Git GitHub. D
- 14:34stands for debugging. Debug with care.
- 14:37Debug as you build. And C stands for
- 14:39context. Provide as much context as
- 14:42possible to your AI coding tool in the
- 14:44form of screenshots, examples, more
- 14:47detailed descriptions. The more context
- 14:49and details that you can provide your
- 14:51AI, the better able it would be able to
- 14:52build what you have in mind. As a bonus
- 14:54tip, if the hyper-specific AI app that
- 14:57you're trying to build, you also want to
- 14:58include an AI component to it, I know,
- 15:01right? Very meta here. You're using AI
- 15:03to build an app that has AI in it. Think
- 15:05carefully about what kind of model you
- 15:07want to use. I'm going to link in the
- 15:08description a couple of videos that I've
- 15:10done that may be able to help you decide
- 15:12which models to use. Wow, great. If you
- 15:14follow these steps, by now you should
- 15:15have the first version of your
- 15:17hyper-specific app. The first time that
- 15:19you do this probably going to take you a
- 15:20little bit longer, especially if you've
- 15:22never done AI coding or built apps
- 15:24before. But, as you build more and more
- 15:26of these, I promise you it's going to
- 15:27become so much easier and faster. These
- 15:29days, depending on how complex the app
- 15:31I'm trying to build is, I can usually
- 15:32get the first version out in maybe like
- 15:352 to 3 hours up to like a day. Just
- 15:37saying, prior to AI coding tools, the
- 15:40development of really any type of
- 15:41application would have taken you like
- 15:43months to do. And there's no way you
- 15:46could do this if you didn't know how to
- 15:47code, obviously. But, now you can, and
- 15:49you can do it in like a few hours. That
- 15:51is pretty magical. All right, time for
- 15:53the last step of this framework in
- 15:55developing your hyper-specific app. And
- 15:57that is hosting. Hosting is providing
- 16:00the infrastructure of where your app
- 16:02lives and runs, so you and other people
- 16:05can access it. Roughly speaking, there
- 16:06are three different categories of
- 16:08hosting options. The first one is cloud
- 16:10options. This is when you can rent
- 16:12compute, and you can dump this code like
- 16:14the application that you build and have
- 16:15it run on these cloud services, and like
- 16:18get other people to run it for you. You
- 16:20pay for how much compute, like how much
- 16:21running of this application that you
- 16:23use. It can scale better, so you can
- 16:25have one user using it or 10,000 people
- 16:27using it. And there's also other tons of
- 16:28managed services that you can use as
- 16:30well, like databases or queries. As an
- 16:32analogy, sort of like renting a hotel
- 16:34room. You're paying per night or hourly,
- 16:36sometimes. There's like managed services
- 16:38like a concierge, there's like a mini
- 16:40fridge, and you can like invite lots of
- 16:42other people over, and you can like
- 16:43leave anytime you want as well. You
- 16:45don't have to like clean up after
- 16:46yourself. This might be the best option
- 16:48for you if your hyper-specific app that
- 16:49you're building you can foresee that
- 16:51there's going to be unpredictable
- 16:52traffic. There might be like a lot of
- 16:53people using it at the same time, and
- 16:55then suddenly like nobody using it for a
- 16:57while. Or they involve like other team
- 16:58members coming in and also like tweaking
- 17:00the app and doing stuff to it as well.
- 17:02And any that involves having to scale
- 17:03and you don't really want to deal with
- 17:04the complexities of that. The downside
- 17:06of this is that it can get very
- 17:08expensive cuz it's like a hotel room,
- 17:09right? You get like all these like fancy
- 17:11things as a premium option here. You may
- 17:13be locked into like a specific vendor
- 17:15and their services, and it's not
- 17:17private. What happens in the hotel
- 17:19doesn't stay in the hotel, potentially.
- 17:21Some examples of a cloud options would
- 17:23be like AWS, Google Cloud, Azure. I'm
- 17:25going to put on screen now some other
- 17:26examples of the cloud option as well.
- 17:28Now, the second option you can consider
- 17:30when you're hosting your hyper-specific
- 17:32app is a VPS, a virtual private server.
- 17:35This is a fixed virtual machine that
- 17:37you're renting out 24/7 for a flat fee.
- 17:39It's kind of like renting out an
- 17:41apartment. You have a fixed monthly
- 17:43amount that you're renting it for and
- 17:45during that time it's your
- 17:46responsibility to keep it tidy and not
- 17:48disturb your neighbors. And you can also
- 17:50like decorate it and do what you want to
- 17:51it within reason. This is good if you're
- 17:53going to be building a bunch of
- 17:54different hyper-specific apps that are
- 17:57all like relatively low traffic, as in
- 17:59you're going to be the only person using
- 18:01the app or maybe just like a couple
- 18:03other people. It's customizable, you
- 18:04manage it yourself, and you have a
- 18:06little bit more privacy as well cuz what
- 18:07you store inside, like data for example,
- 18:10is a little bit more private. Do you
- 18:11want to make a caveat here that's not
- 18:13completely private though because
- 18:14technically your virtual machine is
- 18:16still physically sitting on somebody
- 18:19else's data server, right? Like the
- 18:20hardware still belongs to somebody else.
- 18:22So, if they like really, really, really
- 18:24wanted to go take a look, like your
- 18:25landlord if they really, really, really
- 18:26want to come in and look at your
- 18:27apartment, they can. They still have the
- 18:29keys. Like they probably won't, but they
- 18:31could. So, yes, that is one of the cons
- 18:33and you do need to manage things
- 18:34yourself. You got to clean up after
- 18:35yourself. And even if you don't have the
- 18:37app running, you still have to pay for
- 18:38that rental fee. So, I think this is
- 18:40more worth it if you do have multiple
- 18:41apps that are going to be running or you
- 18:43are actually going to be like constantly
- 18:45using the apps that you built. Some
- 18:46examples of VPS providers include
- 18:48Hostinger, Hetzner, Digital Ocean. Put
- 18:50some other examples on screen now. Now,
- 18:52finally, the third option is hosting
- 18:55your hyper-specific app on your own
- 18:57hardware. This is a physical machine
- 18:59that you would just buy one time and it
- 19:01would just like run on your desk.
- 19:03Continuing with our property analogy,
- 19:04this is like truly owning a home. The
- 19:06pros like you buy it once, it's truly
- 19:08yours. You really have like complete
- 19:10privacy. Nothing is going to leave your
- 19:12local machine, and you exactly know what
- 19:15you want to do with it. Fully
- 19:16customizable to whatever you want it to
- 19:18do. And after you buy it one time, the
- 19:20only thing that you really need to pay
- 19:21for is like just the electricity for
- 19:23running your hardware. But, just like
- 19:25owning a home, there are also downsides.
- 19:27Because you own everything, you also are
- 19:29responsible for everything. If something
- 19:31breaks, you got to fix it yourself,
- 19:33completely. Something needs updates, got
- 19:35to do it yourself. Your home internet
- 19:37could also be a bottleneck if it's not
- 19:39fast enough. Physical failures,
- 19:41downtime, things like that. This
- 19:42technically could also be the cheapest
- 19:44option for you if your laptop is
- 19:46relatively modern, because you can
- 19:48actually use your laptop to be running
- 19:50your app. But, the issue is that if
- 19:52you're like doing other stuff
- 19:53simultaneously, it might make your app
- 19:55really slow or just like completely
- 19:56shuts down. Or if you close your laptop,
- 19:59then it would stop working. So, that's
- 20:00why people often choose to buy separate
- 20:02hardware to run their apps. Like, that
- 20:04machine is going to be dedicated to
- 20:06running your hyper-specific app, and
- 20:08it's going to be running 24/7. The very
- 20:10popular option is to buy a Mac mini for
- 20:12this, but I'm going to put on screen now
- 20:13some other options that are available as
- 20:15well. Cool. Now that you've have hosting
- 20:18done, you officially have a properly
- 20:20functioning hyper-specific app. Isn't
- 20:22that cool? I'm genuinely so excited to
- 20:24have been able to share this this
- 20:25workflow with you guys. I use it so
- 20:27much, and it really is my favorite AI
- 20:30workflow right now. I hope you're going
- 20:31to be building your own hyper-specific
- 20:33apps soon, too, if you haven't already.
- 20:35Let me know in the comments what you
- 20:36want to build now. Thank you so much for
- 20:38watching until the end of this video,
- 20:39and I will see you guys in the next
- 20:41video or live stream.
About this transcript
This page contains the full transcript of My Favourite AI Workflow Nobody Talks About by Tina Huang, generated from the public captions YouTube serves with the video. The transcript has 4,847 words across 693 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.