Code Riff

We vibecoded an app for our grandparents

• Eric Tan, Yaohong Ch'ng • Episode 7

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 36:14

This is a secret episode we almost didn't release. Eric (vibe-coder, can't read code) and his cousin Jason (a software engineer) sat down to build a family app for their ageing grandparents. What you'll hear is a distilled 2.5h chat, with a brief demo near the end.

We're releasing it because the mess is the whole point. If you're curious about vibe coding but scared to start, this is what it actually sounds like when two real people try.

Key takeaways:
- Get the AI to grill you before you build. Jason had it question his intent and features until they reached shared understanding - no code until then.
- Leave the technical details out of planning. Focus on the job to be done and who's using it; the stack decisions come later.
- Planning is painful but it pays off. An hour of arguing and changing our minds upfront beats iterating ten times on the wrong prototype.
- A software engineer meets Claude/Opus for the first time and reacts to how it works vs his usual Cursor setup.
- You don't need a finished product to start. We ended on a mock app - grab a friend, your cousin, your kids, and just start building.

Hosts: Eric Tan (non-technical builder). Yaohong Ch'ng is the usual co-host - this episode is Eric solo with a guest.
Guest: Jason - software engineer, Eric's cousin, daily Cursor + Sonnet 4.5 user trying Claude/Opus for the first time.

Got a problem you want us to solve live? Fill out the form:
https://forms.gle/DSyLzPAoR6x2M4Np9

A few terms you'll hear:
- Vibe coding: building software by telling an AI what you want in plain English, instead of writing code yourself.
- Claude / Opus / Sonnet: Claude is the AI assistant; Opus is the bigger, pricier model Eric uses, Sonnet the faster, cheaper one Jason uses at work.
- Cursor: a code editor with AI built in - Jason's daily tool.
- Plan mode: a Claude Code setting where it asks you questions and writes a plan before building anything.
- Notion as a database: using Notion to store the app's data instead of paying for a proper database - a cheap shortcut with limits.
- Mock data: fake placeholder content used to see how the app looks before real data goes in.

Connect with us:
- Eric on LinkedIn: https://www.linkedin.com/in/erictisme/
- Jason on LinkedIn: https://www.linkedin.com/in/ysjasonchen/
- Email: code.riffs.ai@gmail.com

Code Riff - messy real-world problems, solved with AI, so you can too.

Speaker 6:

Hi guys. Eric here. Just a quick word that this episode was recorded about one and a half months ago. That said a lot has changed. Some of the prompts have become skills. People are moving into coding apps like the Claude app and the Codex app. Anthropic, also released their new model Opus 4.7. But I want to assure you that whatever you see in this episode would still be applicable, especially if you're still just starting in your journey coding with AI or learning how to develop new apps. So I hope this episode inspires you to grab a friend or a family member or even yourself and get coding because it's never been easier to build.

Jason:

I spend a lot of time planning my, my features actually, I spend like maybe like. Half an hour to maybe two hours, like just writing an entire prompt.

Eric:

We are like fighting on screen. Oh, no. Oh no. Restore code, and conversation.

Speaker:

We managed to get a mock app up, which would have taken Jason A. Day or two a day to make maybe

Speaker 2:

definitely manually, probably more than a few days.

Speaker:

Yeah. And we did it in like two and a half hours.

Eric:

Hi everyone. Welcome back to the Code Riff podcast. I'm Eric Vibe coder, and today with me I have Jason, who is a software engineer

Jason:

i'm Eric's cousin.

Eric:

Why Jason came over is because he wanted to build something fun for the family, so that our family members can hang out with each other more easily. Especially our grandparents who are also getting a bit older. So We'd like to vibecode, something that would be able to be useful for helping them hang out with each other. Find new places, you know, things like that. So it's going to be quite free flow. Jason sent me a few voice messages yesterday, Jason, do you mind just summarizing what you, what is the initial thought.

Jason:

Yeah. Okay. So, my grandparents are getting old and basically they don't, they don't have a lot of things to do, right? Very old people. Like, they don't do a lot of things and then like, they don't do a lot of things. Then they'll just start to like, kind of like, deteriorate at home. Like they don't, their brains are not stimulated, their bodies don't move and like they're a bit less excited about life. And I think when that happens, then that's when you start to get old and start to get sick and you start to, you know, bad things that happen. So like I was thinking, how can I stimulate or engage them more? And I was thinking of certain pain points that I felt about, maybe they don't feel it, but I just feel like maybe I could make some kind of an application to make it easier for. Us to meet up as a family, as well as keep track of where we have eaten before as a family. Like an easy place to check. Because sometimes we go to places with my grandparents and we eat stuff and we forgot that we've been to a place. And then when we go back we kind of forget like whether the food was good.'cause my, my grandfather likes to write reviews about food places. So firstly we are a bit lazy to read it. Secondly, like, it's also like in Chinese, so like maybe the younger ones struggle to read it a bit. So yeah, I thought that it, it would be nice if we have an app that can kind of like help everyone see where the family's eating. Even create like open jio so it's like easy to invite people and whatnot. I mean, now it's pretty easy. We use like WhatsApp and all, but you know, that's just an additional feature. I guess the main feature, it would be something like a map we can keep track of where we've eaten before and an easy place that we can see all the reviews that, that my grandfather has written, you know, so it is like kind of easier for us to keep track of. The reviews as well as to having one place so that we can easily access it. Yeah.

Eric:

let me just show you what Jason's talking about in terms of the essays that my grandpa writes. This is a notion, and he writes about places that he's been to in Malaysia, you know, and also some of the favorite foods that he has tried in, in even Singapore. There are many restaurants here, maybe even like 50 articles or something.

Jason:

Maybe I can talk about like my motivations real quick.

Eric:

Okay.

Jason:

Yeah. So actually I have two motivations to embark on this project with Eric. Yeah. So firstly is basically yes to create an interesting app that maybe my family might use. So whether they use it or not, not very important. So. Just wanted to try make something over a weekend, see how good like coding has become with ai, you know? So that's the first motivation.

Eric:

Yeah.

Jason:

The second motivation is basically, so Eric is an avid Claude user and I actually haven't used Claude before. Mm. I've been using Cursor mainly at my work and for my personal projects. So I kind of wanna just see how Eric uses Claude, you know, learn something about claude, see like whether you can win me over as like a user. Yeah. And just like learn about the differences between Claude and like cursor. See what was the big differences and see how far Claude can kind of take my ideas and run with it and go crazy. Yeah. I wanna see how, how good Claude can take my ideas and put it into real life.

Eric:

Yeah. Thanks. Thanks Jason. So what model is your favorite right now?

Jason:

So usually at work I use, sonnet 4.5, like thought. So it's pretty good. It does most coding tasks very well. It understands, it doesn't hallucinate that much.

Eric:

Yeah.

Jason:

So, yeah, I mainly use that at my work and I guess. In my free time as well as for daily usage, I use Gemini and GPT. Right. So they're, they're easy to use, good interfaces, and they're sufficient for like daily search and all. But yeah, specifically for coding as well as like product feature creation and iteration going end to end from like, planning to like coding to even like deploying it to live. Yeah. I wanna see how far this can take us.

Eric:

Yeah. Okay, cool. Actually right now I am using Sonnet 4.6 and Opus 4.6. So we are one generation above in terms of what you get to use in work.

Jason:

Yeah.

Eric:

I guess because your company is also trying to control cost, right? When it comes to using tokens, which is probably why you're not using to Opus. Is that why or

Jason:

no, actually. I've just been used to using 4.5. So I, I, I've kind of never found,'cause I, I, I see Opus in the selection.

Eric:

Okay.

Jason:

And I've just never really got to know Opus. You know, like they, they, they kind of, deploy new models all the time. And, and it's like, you, it's like working with someone new every time I'm using a new model. Right. It's like, kind of like, like why do I wanna take a risk if a new intern when my current intern is doing fine?

Eric:

I see.

Jason:

Yeah. So I, I've used Opus once or twice and it, it looks like it's functioning totally differently. Like, it's like a new supercharged intern.

Eric:

Yeah.

Jason:

Yeah. So I, I'm still working my way to using it like more, I guess.

Eric:

Okay. I'm very curious to see what you think about Opus today. Okay. Because that's what we are mainly gonna be using.

Jason:

Very cool. Very

Eric:

cool. And I personally find Opus very good at understanding intent. So you might not need to be extremely detailed about what you tell it.

Jason:

Okay. And

Eric:

you will try to figure it out.

Jason:

Okay.

Eric:

Perfect. So that's why I guess as a beginner in coding, I use Opus. where would you start?

Jason:

I usually like to go from scratch because I like to, I like to vet the, the feature requirements, so usually I spend a lot of time planning my, my features actually, I spend like maybe like. Half an hour to maybe two hours, like just writing an entire prompt. And I should go back and forth with the AI to actually ask me like, what am I missing? What do you think I should focus on? What do you think is not so important or not so useful? And question me about my intent. You know, like, are we spending too much time or resource on a certain feature before I actually start? So I actually prefer solid plan before I start.

Speaker 3:

Mm.

Jason:

Yeah. So this is, this is just to a contingency, right? In case it things go wrong. So I wanna make sure that I spend less time later reiterating, I wanna make sure that the, the first prototype that comes out

Speaker 3:

Yeah.

Jason:

Is gonna be something at least nearer to what I want. Then something that I have a prototype, but I need to iterate like five, 10 times before it starts to get useful.

Eric:

Yeah.

Jason:

Yeah. What do you think?

Eric:

So, okay. So let's, let's plan. I'm Vibe Coding with Jason and Eric. Yesterday, Jason or so wrote out something, right? That's quite important.

Jason:

Yeah, that one is like more closest to my intent, you know,

Eric:

which is,

Jason:

I guess you can copy in the, the second part line. The first part also important, the features part.

Eric:

So Jason spent,

Jason:

you can add in this

Eric:

feature some time writing this. paste it fine. I think I'll just copy everything.

Jason:

yeah, sure. It's fine too.

Eric:

Spend some time thinking through. Okay. And then on top here, it's just like a brief motivation as well. And then Jason also said, spend some time thinking through. And I'll add that, the goal of this is also for the family to come together in a, a meaningful way that helps everyone find the right time to come together and have a nice variety of meals with our grandparents. and also longer term wise, it's to, think about yeah, having a good mix of food and also, one crazy idea I have is also like to think about how we can make the meals themselves better, or more meaningful with conversations. So that's just one additional point that I would add, to dilute Jason's intent.

Jason:

So maybe, you help me add one more line underneath. So usually what I do is just, so let's not include any, technical, details. And, uh, question me about the, the features and, give me questions about what you think it's important about this and what do I want the features to look like. Yeah.

Eric:

Yeah. Very, very cool. Very cool. I've never tried this before, you know, not including any technical details. That's very interesting.'cause you basically wanna get to the job to be done. Yeah.

Jason:

Yeah. Okay. So add like, uh, yeah, let's focus on getting the requirements right as right as, as it can be for now. So yeah, I want you to really grill me on what this app is gonna look like, the user experience, who's using it, and what the app, entire app experience is gonna be like. Whether it's gonna be a website and a web application, phone application. yeah. And the call features.

Eric:

So one thing I have to add here, plan mode actually activates a skill called AskUserQuestions.

Jason:

Okay.

Eric:

So it's meant to interview the user already.

Jason:

Okay.

Eric:

so that's, that's interesting. I mean, we can always do, do that to make it more robust, I guess.

Jason:

Yeah, yeah. Oh,

Eric:

okay. but at one point, and, and something else I would add is that I would like to use Notion as a database, because I don't wanna spend money to make another supabase base, project for this.

Jason:

Yeah.

Eric:

Okay. Are you okay?

Jason:

Yeah. I guess you can just know that we wanna use Notion as a database because it's, we are not discussing anything technical now. Right. So Yeah, it's, doesn't any matter now we're just going through the requirements. Yeah, I think it's fine. You can just start the plan and see what comes up

Eric:

So it has a round one of questions.

Jason:

Okay. I guess I can answer all these.

Eric:

It's asking us very nicely. MCQ, very nice. Who will use this extended? Okay. Extended family. 10, 25 people.

Jason:

Next

Eric:

language. How should the app handle language?

Jason:

I

Eric:

think it should be bilingual.

Jason:

Both. yes. Bilingual and easily like, switchable, perhaps

Eric:

bilingual and easily switchable.

Jason:

Yeah.

Eric:

With toggle

Jason:

for Young and Old.

Eric:

For young and old to use. Right.

Jason:

Let's go next.

Eric:

I wanna say, take note. make UX extremely, elderly friendly.

Jason:

Okay.

Eric:

Yeah. Platform. How should people access the app?

Jason:

let's do it. website first is fine. Website first app later. Yeah.

Eric:

Okay.

Jason:

Does everyone need their own login? No.

Eric:

Shared access, no login, but maybe you have a password. Okay. But just so that no one hacks and reads all our family chats.

Jason:

Yeah.

Eric:

But have a secret password that we can. Share privately go.

Jason:

Right. Great. Okay. So, okay. The Met review is great. Next, more questions.

Eric:

When someone opens the app for the first time, what should they see? What is the home screen map? Um,

Jason:

what do you think? I don't know, like, I'm imagining in my head like, kind of like a mixture of a map. Okay. Yeah. A mixture of a map and like some feed of activities or like recent meals kind of idea.

Eric:

Oh wow. It's showing us three different designs.

Jason:

Oh, that's what it yeah. Maybe something like, you know, you know, Google Maps has this thing where you have like. You have this bar below, you can swipe up. Then maybe this bar, you can have the recent - the feed. You know, so the main thing that comes out is the map, right? But then you actually have this bar to see like the recent like thing, you know, something like that. Or even it could be like half the screen like that. What do you think? So something like

Eric:

this, I think we shouldn't do that because I think Gong Gong and mama wouldn't be able to swipe a tab up.

Jason:

Okay.

Eric:

It should be like one page and then you just scroll down. But if you want to, so I think the dashboard mix it good. Of course I wanna improve it a bit as well.

Jason:

But then in that case then maybe just the map is fine, then the dashboard can just be another tab underneath. Then there's kind of no need to mix it together, I guess, to make it simpler for them to use. Because in this case then,

Eric:

but actually the

Jason:

map is gonna be really small.

Eric:

What we said, we don't know whether they know how to use the map.

Jason:

True.

Eric:

So map is not very user friendly for elderly.

Speaker 3:

True, true, true.

Jason:

We can chat about this. We, we, we can, we can leave this out for now and, you know, we can chat about this.

Eric:

Okay, now we have to chat about this.

Jason:

so okay, let me re ask the remaining questions. Hosts. The home screen focus on, I think Gong gong's articles are fine. What do you think?

Eric:

Yeah, actually I like the map. But I think what makes more sense is for Gong Gong's articles to be repurposed and showcased.

Jason:

Yeah.

Eric:

You know, like highlighted.

Jason:

Yeah. Yeah. And then the map is just another tab where people can go for, to just check where they've eaten before or like find a specific review of a certain restaurant somewhere in the map so they know like where it is, you know?

Eric:

Yeah. The map is for us, the articles are for, for our grandpa.

Jason:

Yeah.

Eric:

And for us to engage deeper with his articles.

Jason:

Yeah. So I guess you can just click first one.

Eric:

Okay, fine. For reactions and comments, what level reactions? Only short comments.

Jason:

let's, let's keep this for V one. I think it's fine.

Eric:

No, I think reactions plus short comments is fine.

Jason:

I feel like we can always add this later if we find like the app is useful, people like wanna add, comment and stuff. sure. Bit beyond Singapore. I think Singapore's fine. Only in, for now is fine.

Eric:

I think there were some places in Malaysia though, in kl and a lot of places in KL actually.

Jason:

But it's gonna be like 80, 90% Singapore. So,

Eric:

and Gong Gong is gonna write one on China, so I, I would like everywhere. Anywhere, but then

Jason:

Okay. Any, anywhere. But the maps focus should be in Singapore because I, I don't know whether we're using Google Maps, right. Or we are just gonna be a manual. They're gonna render just a map of Singapore and then we're just gonna have pins around

Eric:

Anywhere in the world, but focused on Singapore. So each time we open an app, it should be Singapore. Of course, there are articles in kl. And parts of China, but let's focus on Singapore always.

Jason:

And not every article has to map to a restaurant because we sometimes don't have the pins or the links to every single restaurant we eat in.

Eric:

All right, let's go.

Jason:

So do you really spend so much time planning? What do you, what do you think?

Eric:

Nope.

Jason:

Okay. I guess then this is a new experience

Eric:

for you. No, I think sometimes I do it, but yeah.

Jason:

Okay.

Eric:

It's just that recently what I do is I generate a prototype and I click around.

Jason:

Okay.

Eric:

And then I see what's missing. So

Jason:

you iterate on it.

Eric:

So as I look at what I meet, I continue iterating.

Speaker 3:

Okay.

Eric:

Yeah. Okay. How does Gong add a new article? He writes an Apple Notes. What's the easiest way to get into the app? Share in Apple Notes and it goes into the app.

Speaker 3:

Wait, wait.

Eric:

Very complex, but I, I don't wanna, I mean, this solves my pain point.

Jason:

No, there's no different from pasting into a form. Right. Let's say, let's say the app has a map, has a form that he can just type this in. He doesn't need to type it in Apple Maps. No. It, but

Eric:

he loves Apple Notes. Like have, have you used Apple Notes? It's very nice.

Jason:

Okay. Okay. He loves Apple Notes. I see.

Eric:

Sorry, I am going to. Do Apple Notes for now because that would be very amazing. Okay. Like it would save me a lot of time.

Jason:

Okay. Yeah, yeah, yeah. We can go ahead there. I'll just be skeptical on how it's gonna be done, but,

Eric:

okay. Thank you. Jason, when someone clicks the WhatsApp link, what should they see? One is a simple RSVP card. Oh, this is quite nice. It shows the place, the time. Who's going or the RSVP restaurant info. Oh my gosh, this is cute. I, I actually like this, but it is not as practical as well,

Jason:

I think a simple card is fine. Yeah. Simple card is fine.

Eric:

We need a purpose.

Jason:

we purpose lets comments, less comments.

Eric:

No, I prefer why you should go. It's like more conversational.

Jason:

Why you should come. Okay. Why you should come.

Eric:

Thank you Jason. We are like, we are like fighting on screen.

Jason:

Okay, cool. Cool, cool, cool. Cool. Cool. Cool,

Eric:

cool, cool. Okay. Enter, submit. Ah, I need a stretch.

Jason:

cool, cool, cool. I'm excited. So generally, you don't need to know about the, the technical details, right? As a non-technical person, like how, how do you know you're making a correct decision?

Eric:

I need, I usually wanna know the technical decision so that I can get the right API keys ready. Honestly, that's all the work for non-technical coder.

Jason:

Yeah. I see.

Eric:

So I just wanna make sure that everything is functional and how it's functional. It's for me to go and grab the keys. Yeah. Actually just main thing and, and yeah, I really don't need the technical stuff. All right. We have a plan. Okay. Here's our table. Product requirements, context, grandparents getting older, we need more family oriented activities. grandpa writes food articles. No one reads them. Family eats out regularly, but forgets where they've been. Article output has been declining 'cause grandpa doesn't feel people are reading.

Jason:

Question Eric, because you mentioned the database is Notion, right? But now that if you wanna directly upload to from Apple Notes, like, you know how, how's that gonna work?

Eric:

Oh yeah. Because notion might be good for just a table, a simple table, but it not, might not be able to store images. That's what I saying, like what was your concern then?

Jason:

Currently, how do you, so you store the entire article inside Notion. Even with the photos, is it?

Eric:

Yeah, I upload the Apple notes. I, I copy paste it in the notion, and then I upload the photos one by one.'cause the photos cannot be copy pasted.

Jason:

Oh, can I see your notion?

Eric:

I mean, Gong Gong has to share with me the, the note first into my phone, and then I upload his notion, oh, I look shabbier. Yeah. Okay.

Jason:

Yeah. You've lost some weight, Eric. Yeah. Let's let, let's not worry too much. Whatever.

Eric:

Yeah. For

Jason:

now,

Eric:

you just need to have a bit of like irrational optimism sometimes. Yeah. And then you see what works

Jason:

possible, possible

Eric:

and then you get disappointed and then you know what is possible. Yeah. I think it's good to always know what it's at the edge of what's possible. Okay. The goal is a simple, delightful app that brings, number one, brings people together. Number two gives the writings a spotlight. And number three becomes a long-term family food memory log. Okay. The navigation is three bottom tabs, articles, map and jio. Okay. Actually that sounds quite elderly friendly. If it's very big on the bottom, it's quite easy to click. Okay. Feature number one is articles. So this is the default screen articles are displayed. There is a full article view and then there is a way to input for our grandpa. So feature two is map and then there's a restaurant card with the restaurant photos. Three bullet summary. Okay. And a button to read the review from our grandpa, okay. Feature three is jio for meal invitations. Simple flow for organizing family meals. Not where people discover meals. WhatsApp is the main channel. This makes it easy to create and share invites. So this is the jio flow. Number one, plan a meal. Number two, pick a restaurant. Number three, set a date and time. Number four, right where you should come. Number five, have a shareable WhatsApp message with a link. Number six, copy and send to the WhatsApp group that they want. Okay, the RSVP page. After you've clicked into the link, opens a app browser, shows the restaurant the time where it should come. Who's going? I'm joining button. And then after joining, your name appears so there are no accounts. Yeah. Person taps, I'm joining and types their name. Perfect. Perfectly simple. Okay, Joe Tap is a list of upcoming meals, past meals, plan a meal. The meal. Anyone can upload photos becomes part of the restaurant visit. UX principles, elderly friendly, bilingual, minimal friction, WhatsApp native and spotlight of the articles, technical approach, notion database structure. And then there's a built phases, phase one foundation, phase two articles, phase three, map, phase four, jio phase five photos and polish. And then there's a verification to test everything's working.

Jason:

actually I was thinking for the, a little extra feature is just that, you know, we each have like a, we have like a photo. Inside the app, like our face photo.

Eric:

Yeah.

Jason:

So easy to see who and for which meal. We just see the, the bubble of our faces and not like the names of who. And that was like my idea originally. If I click into a visit, then I just see a big on top of just the faces of who and you know, and like me as small name underneath, just to make it a little more cute.

Eric:

Okay.

Jason:

So that's my original idea.

Eric:

Okay. So we should implement another feature where in the settings of the whole app, since we have no accounts, we should be able to write a person's name and then upload a photo, a profile picture for each of the names. And then when we are writing, who is going for the meals, as we type the name, the a dropdown of names should appear. And then when you click the right name. It should show the name. And then ideally it should show the profile picture with a name below the profile picture.

Jason:

Just click on the faces

Eric:

or even clicking on faces.

Jason:

You can just, so like, I'd

Eric:

rather search rather than click the 16. It's not all 16 faces will come out at once, you know,

Jason:

but searching 16 phases is very easy. If I have a, like a, like a two columns of just faces, it's just eight rows. And it's very easy to click eight rows of

Eric:

big

Jason:

phases.

Eric:

And if I were to riff on that, these phases would be quite translucent, transparent. But when you click it, it becomes opaque, right? Which. A user friendly way of indicating who is going.

Jason:

Yeah, like the background of, of it is like, each of them is gonna be on a card and when you click on it, it's just gonna light up as green. Okay. Then you can click like confirm to create the invite.

Eric:

Okay. Here's where I have to always decide whether I clear context or continue.

Jason:

Yeah, yeah. Sure, sure, sure. I have no preference.

Eric:

Okay.

Jason:

Claude has written a plan, would you like to proceed? What is clear context?

Eric:

Claude, I like the plan. we also need to create an intent.md where it would capture a lot of the intent behind why we are building what we are building. And then we also need to create a/savechat, basically containing the initial, original prompts that we had. So we can always go back to that because this chat that we just had is extremely valuable and we don't wanna lose any of the original meaning that came out from our, our mouths. as you have already summarized a lot of things. So I think we do need to create those files and remain faithful to what we wanted. And also that helps us to plan for other features if needed in the future. Yeah. What do you think of this? Do you have to do this in Cursor or, I mean, I really, I'm still struggling with this step, the final step. That's why I do a Ralph (loop), because what the Ralph does is it makes like 60 features or less. And then I know that it has captured all the features that I needed in a very long thing that it immediately executes. But how do you manage long, long stuff and so many things you have to, you put it in linear, is it, or Jira,

Jason:

or long things are. So I struggle with that Also, like keeping important context. So usually I, I just try to summarize things. Okay. I'll ask Cursor to put it into a markdown file for later. So now even usually creates its own file and it finds that something's important, but I find it creating too many files. So I, I still struggle with that. Like keeping important context somewhere.

Speaker 3:

Mm-hmm.

Jason:

Yeah. But then again, I guess important context should then be in some document beforehand and it shouldn't kind of be like an impromptu thing that appears in the middle of a chat, you know?

Eric:

Yeah,

Jason:

yeah. But I guess we are vibe coding and you know, usually after an hour and a half we've only set up our environment and stuff, but yeah, now we can spend more time on figuring out what we want instead.

Eric:

Oh, I still using notion. I still thinking about Notion. Should we challenge that? I think we should challenge that. Right? Do you have any comments on that?

Jason:

Yeah, yeah, yeah. As in,

Eric:

okay, let's, we'll do one more, one more,

Jason:

because I mean, yeah, I noticed because I'm coming from a technical standpoint, like how the heck do you like, link Apple notes to like notion to like your application? It is generally just a little bit like spaghetti, a bit messy. And yeah, it should come off with like a red reflect, like, like straight away. Like how is that architecture gonna work? It's definitely something we are, is it gonna just take some shortcut to do some stupid feature that might not work? You know, so that's where I start thinking like, how do we link these things behind the scene? So usually for this technical details, I also prompt the, I prompt the model to ask me.

Eric:

Can you just prompt now?

Jason:

Okay, so like, how are you gonna handle the data source? How are we linking Apple notes to notion to our application?'cause currently we already have Notion, which is just, we have a folder with many pages of articles that my grandfather has. Yeah. How, how are we gonna link this all together? Have you considered these?

Eric:

it is possible to use Apple Notes send it to Notion.

Jason:

just images only external new external URLs.

Eric:

Yeah. That's why you can't just like copy paste from Apple to Notion.

Jason:

So either way it's gonna be a problem for photos, right?

Eric:

Yeah.

Jason:

Take photo, upload image.

Eric:

so maybe Eric should do this manually first, right? I assume we should kill this feature.

Jason:

Yeah, let's kill the upload feature. Automating upload feature. And Eric can just upload to Notion as it is now. Yeah. Then it's just gonna scrape from your notion somehow. Yeah. I guess this is a feature we can try to iterate on later if it, if people use it. Mm-hmm.

Speaker 3:

Maybe uploads to notion,

Jason:

so when it, when I add a photo to Apple, it's basically just reading from external link. Like Apple just saves in some Claude and it's a link.

Eric:

Yeah, it's a link. So when you copy paste, adding think notion, uploads onto it, servers,

Jason:

this is what you do. Like when you stare at the screen too much, you change your wallpaper. Ooh, this is nice. But yeah, I do find myself staring at my chat in cursor for like really extended periods of time and it's, it's actually quite exhausting.

Eric:

You get eye strain, you know, after 20 minutes you're supposed to look very far for 20, at least 20 seconds.

Jason:

Is it

Eric:

every 20 minutes? And my eyesight has actually got worse. Yeah.'cause I think I stare quite long on my computer.

Jason:

Oh no, Eric.

Eric:

Yeah.

Jason:

Okay. What do we have now?

Eric:

Okay. very quickly, the product, new product requirements.

Jason:

Let's skip to the part where we deleted.

Eric:

Shall we just go?

Jason:

Yeah, yeah, yeah.

Eric:

Clear context and bypass permissions.

Jason:

Is that what you really do? Do clear context? Does it make it better? I,

Eric:

up

Jason:

to you.

Eric:

I think it's good. Just so like now this is the canonical thing.

Jason:

Okay.

Eric:

We spend the past hour and a half like making this thing. So it's just gonna follow this thing.

Jason:

Okay, cool. Let's go ahead.

Eric:

Okay. And then we also want it, we also wanna see how it works and explain how it works, all that kinda stuff.

Jason:

Okay. Yeah. But I, I also face this frustration where I spend too long already and I just wanna keep going. I, I just like, I'm so sick of it. Just start.

Eric:

Yeah.

Jason:

Yeah. And I, I think that kind of like, kind of decreases the output over time, like

Eric:

Yeah.

Jason:

Since like, you know, you're really interested at the start and then we are really detailed and then we just get more frustrated and we just wanna do, have something to play with.

Eric:

Yeah.

Jason:

Yeah. I face that like all the time, like mm-hmm. Work in my personal projects. I spend like an hour planning and then the quality just goes down and down and I just wanna start because of that. I end up with something shitier than if I just spend more time planning.

Eric:

But when you say shitier, it is also compared to you're comparing it to a state where you have a mental, like, kind of infinite mental capacity to plan out all these crazy features that you are doing. In that amount of time. Right.'cause right now the output is really tremendous. But what you're comparing to is like a tremendous amount of output with, uh, sustained like mental capacity.

Jason:

Yeah.

Eric:

And you're comparing it to like something quite unreasonable, in a sense.

Jason:

Yeah, that's true. Even because we're speed running it. So like we, the battery gets drain faster even for our own brains.

Eric:

Yeah. You need more

Jason:

water? Yes.

Eric:

All right. I just took an eye break so it didn't, okay. It saved the plan and then it's doing the self review. What is this? So it wrote the plan. Oh, so it continued to write the plan using a skill to write the plan. It looked into v1. Oh, no. Oh no. Restore code, and conversation. How long has it been? Two hours. People actually say they miss, oh, 4.5 I, wow. This is exactly what you're talking about with Sonet. Like you, you're used to sonet already, like people are used to Oh, 4.5.

Jason:

Yeah. Right? Yeah. It's like, it's like a different person and sometimes, you know, you might get a more nerdy, like smarter colleague, but they just have worse social skills and you just don't wanna work with them.

Eric:

4.6 is nice, but spins its wheels forever.

Jason:

Yeah. So,

Eric:

okay. Interesting.

Jason:

Interesting phenomenon.

Eric:

Yeah.

Jason:

How we are describing models.

Speaker:

Okay, so we are done and we lost a lot of, uh, footage. Um, but anyways, we managed to get a mock app up, um, which would have taken Jason A. Day or two a day to make maybe

Speaker 2:

definitely manually, probably more than a few days.

Speaker:

Yeah. And we did it in like two and a half hours. So we do have the app. It's quite nice. Still mock data and we'll see what, how we can put real data on it. So, but yeah, thanks for joining me on this, Jason. Thank you for making this interesting and cute idea for the family. Yeah, I'm

Speaker 2:

sure.

Speaker:

I hope they'll love it.

Speaker 2:

Uh, yeah. I mean, they love it. As long as, you know, we, we try something and we explore AI and you know, you help me understand more about Claude

Speaker:

Yeah,

Speaker 2:

I think it's cool. I learned a lot.

Speaker:

Yeah. All right. Thank you. Thanks Jason. And see you all in the next episode,

Erictisme-6:

Hi everyone. Eric jumping in from the future here. Just wanted to show you what the app looks like after spending a few more hours on it before we launch it to the family. So here you have the main page of the articles that's able to be translated between English and Mandarin. Able to do that within each article as well. With a map page that you're able to look at different places and see what's the associated place with it. And there is also a notes page for random notes from people. There's a planning page where you can organize meals with people and also click into who is joining or who is hosting. So yeah, hope this inspires you to grab a family or friend or yourself to get building.

Eric:

Thanks so much for joining us for this episode, if you found this helpful, feel free to leave us a rating on your favorite podcasting apps or like share, subscribe on YouTube and if you want to get into the next episode, feel free to fill out the form as well. So yeah, thanks so much for joining us. Uh, really appreciate that you even made it to the end. And see you on our next episode. Bye.