編程小白的我是如何用AI做出第一款應用的?

OpenClaw is a hit, and I've set it up so I can send it a message, and it will create a podcast for me and publish it to the internet.

For example, I can ask it to summarize today's AI news, Tesla's earnings report, or even turn the first two chapters of a book into a podcast, or create an audio tutorial from an article.

About ten minutes later, I can open my Apple Podcasts, and an episode made just for me is already there.

This podcast doesn't need platform approval to be published.

It can be a private feed that only I can listen to, or I can let my friends and community subscribe to it, and it's free.

So, how is this possible?

Well, to make this happen, OpenClaw alone isn't enough.

It also needs a private podcast platform that I "Vibe Coded" myself called Dear Tape.

Right, it's like I went to a whole lot of trouble for something very specific.

So in this video, I want to share with you the private podcast platform I made, Dear Tape, and how to use it with OpenClaw to completely automate podcast creation and publishing. First off, why did I even make Dear Tape?

What is it, exactly? Well, it all started with a need I've had for a long time.

For a while now, I've wanted a private podcast.

So, what's a private podcast?

It's... well, I want to create content, but only share it with specific people, like my family, friends, or our community members.

But starting a podcast is actually quite a hassle.

You have to submit it to Spotify or Apple, and sometimes there's an approval process, and all sorts of settings to configure.

The main problem is, once it's live, it has to be public. Anyone can search for it, which defeats my goal of private sharing.

Many years ago, I tried a website called HuffDuffer.

The concept was simple: if you had a Dropbox link, you could create your own podcast.

But the problem was that all your content was public. Anyone could

subscribe to your podcast, so I stopped using it.

But it showed me that it was possible to use storage like Dropbox to host my own podcast for free.

Of course, back then, with no programming background, I never thought I would ever build one myself.

But then AI came along, and an opportunity arose that gave me the idea. Last summer,

my in-laws came to Canada to visit us, and I often drove them out to the Rocky Mountains.

Sometimes we'd be on the road for hours, and I'd think, "Let's listen to something."

Around that time, I had been using NotebookLM to create some audiobook-style podcasts, so I just played them.

It turned out they loved them and started asking me for new episodes every day.

That's when I thought, if only I had a private podcast. My

content could be pushed directly to it, and they could subscribe with the native podcast app on their iPhones to get updates.

No more hassle of sending files or links.

Plus, it would be private between us, no one else could see it. This tool

could also let me create internal podcasts for our community. So, I decided to just do it.

And Dear Tape was born. OK, this is great. This, this

is really awesome!

I was really bored after dinner, so I thought I'd try to see if my private podcast idea could work.

And what's crazy is that in just a few minutes, I had already set up my own private podcast.

You can all have a listen. Yay!

The first version was usable right away. It basically took two shots to solve the problem.

Oh, let me tell you, this Vibe Coding thing lately, it's been keeping me up at night.

I can't even eat properly. All my bandwidth is consumed by it.

Look here, this is the project I most want to do.

I believe... I believe it might be too early to say, but I believe it should go pretty smoothly. But I spoke too soon.

pretty smoothly. But I spoke too soon.

Because I quickly ran into a major problem. Since I was inspired by, uh, HuffDuffer, which uses Dropbox, I used Dropbox too.

But I discovered that Dropbox will ban links that are played too many times.

The link would become invalid after just a few plays, which meant the service would be very unstable.

Of course, I could rent my own server and have people upload audio there, but that would be very expensive, and it wouldn't be free anymore.

My idea was simple: users shouldn't have to pay extra. They just need a free cloud storage account.

So I had to put the project on hold.

But after a while, one day it suddenly occurred to me, what about trying Google Drive?

I gave it a try, and it actually worked. And Google Drive's limits are very generous; you basically don't get banned.

Looking back, it's actually pretty funny. It was like I had this mental block, thinking that only Dropbox could do this, and no other free cloud storage would work. But this also proves that when you hit a dead end, if you just set it aside for a bit and then look at the problem again, you'll often find a breakthrough.

So the project was revived. (Podcast audio) With the help of AI, I spent a few weeks overcoming various challenges, like user logins, databases, privacy protection, and my weakest skill, UI design, and finally, I managed to build the platform.

I named it Dear Tape, and its slogan is "Sound for the ones who care."

The problem it solves is actually very simple. First, it's free.

You just need a free Google Drive account, and you can use it. (Podcast audio) Second, there are no barriers to entry.

You don't need to go through Apple's review process or submit to Spotify.

Just paste in the audio link, and it will automatically generate a podcast subscription link.

Third, it's private. Only people with the subscription link can subscribe. Others

can't find it by searching in podcast apps.

Of course, you can also create, uh, multiple podcast channels, separated by your interests or target audience.

Fourth, no ecosystem lock-in. I think this is very important. Most

subscribers don't need to change their podcast listening habits.

You can use Apple Podcasts and most other clients that support RSS subscriptions to listen to your station.

So what can you use it for?

Uh, for example, a teacher or coach could create an exclusive course podcast for their students, private content, book clubs, internal sharing within communities, or a private radio station for family.

For instance, a dad who often tells his kids bedtime stories has to go on a business trip for a month.

While he's away, he can record episodes and push them to this private station that's just for him and his kids.

I think that would be a really wonderful thing. However, this all sounds great, but in practice, there are two pain points.

First, making a podcast is a creative task, and most people won't do it proactively.

Second, even if you use an AI tool like NotebookLM to generate the podcast, the process of downloading it, uploading it to Google Drive, then logging into Dear Tape to publish it...

that whole workflow is still very tedious.

So after using it for a while, even I stopped.

But then, things took a turn. (Intro music)

You see, I finished making Dear Tape around July 2025, using Replit, an online Vibe Coding platform.

But later I started using local AI development tools on my computer, like Claude Code.

And then I realized something.

And that is, the hardest part is just installing it on your computer.

Once it's installed, it's like you've invited an AI onto your own machine.

It can not only write code, it can help you do everything you can do.

You just need to tell it what you need, and it will find the tools itself, or even write its own tools to get the job done.

And the recently emerged OpenClaw is a similar tool.

Except it's more interesting.

It's like an AI employee living on your computer, but you can assign it tasks remotely through a chat app on your phone.

So I thought, why not let the AI on my computer directly control NotebookLM to generate podcasts, and then automatically publish them to Dear Tape?

To achieve this, I needed three things.

First, an AI that can control the computer, like Claude Code or the more recent OpenClaw.

Second, a tool for NotebookLM. On GitHub,

there's an open-source project called NotebookLM_PY that allows an AI to automatically create notebooks, upload materials, and generate podcasts.

Third, a way for the AI to directly control Dear Tape.

I had the first two, but what about the third one?

Uh, if you've also gotten into Vibe Coding, you definitely know the concept of an API.

You can simply think of it as an interface that lets a program directly call a service.

For example, the voice-to-text tool I made before, Flash Notes, used OpenAI's API.

So I thought, why don't I also make an API for Dear Tape, let the AI call it, and automatically publish podcasts?

So I had the AI help me add this feature.

Now, all three things I needed were ready.

Now every Dear Tape user can get their own API.

Just tell your OpenClaw or Claude Code about the method I mentioned, and it can set up the entire automated workflow for you.

And that brings us to the scene from the beginning of the video. Now for me, whenever I want to listen to a podcast, I just have to tell OpenClaw.

For example, uh, I'm reading a book recently.

I can say, "Hey, turn this book into a podcast for me," and it will automatically handle everything: find the book, split it into chapters, save them to my Obsidian notes, upload it to NotebookLM to generate the podcast, and push it to Apple Podcasts via Dear Tape.

You know what?

When I connected this whole process of podcast generation and publishing, and whenever I want to listen to something, I just have to say a single sentence, and then I can hear a podcast made just for me on my phone...

that feeling is truly magical.

And it genuinely saves me a lot of time.

So, that's my Dear Tape.

Everyone can start using it for free right now to make your own private podcasts.

And there are more features waiting to be discovered. I know that at this point, many of you want a detailed tutorial.

Actually, the easiest way is to just show the content of my video to your AI.

It will help you configure everything step-by-step.

You can also join our new, free community space, which is dedicated to sharing and discussing the use of OpenClaw and Dear Tape.

The link is below the video.

Of course, you can also build your own Dear Tape, completely for your own use.

That was unthinkable in the past, but in the age of AI, if I can do it, so can you.

Often, what stops us isn't the technology, but our own mental blocks.

In fact, Dear Tape isn't the only project I've made.

Over the past few months, I've built a lot of internal tools and some apps that are available for everyone to use.

These apps are installed on my phone right now, and I genuinely use them every day.

It's not a gimmick.

So, I've decided to start a special series on this channel called "My AI Junk," to share the apps I've made and give you all some inspiration.

Dear Tape is the first episode of this series. Of course,

besides sharing my "AI junk," I also have many insights from learning about AI that I hope to share with you.

So I've started a new channel called "Powered by Passion," dedicated to discussing the technical details and thoughts on AI.

I hope you'll follow it.

Of course, if you want to learn AI from scratch and empower yourself, you're also welcome to join our "Powered by Passion" community.

Besides the free section I just mentioned, we also have a large number of subscription courses covering all aspects of Vibe Coding, to help beginners like me get started from zero.

We're even planning to launch a course that teaches you how to replicate Dear Tape.

It will be a very interesting experience, because you won't just get a Dear Tape you can use yourself, more importantly, you'll learn to use AI by building a project hands-on, bring your own ideas to life, and build momentum for your next breakthrough. OK,

that's all for this video.

Thanks for watching.

See you next time.

Bye-bye.