My daughter is heading into 3rd grade. Summer break is nine weeks
Every March or April, parent group chats start filling with summer camp links. Drama camps, language camps, NTU, NTNU's "Little Master" series, KKday, Humanistic Education, Hsin-Yi, Jade Mountain, Eger... two weeks pass and my LINE messages stack faster than I can read them
Then the thought lands: you do AI for a living. Use it on this
Fine. Here we go
The conclusion first: open the tool here
69 camps, 35 organizations, 22 cities. Take a quiz, get recommendations, filter freely, drag into a 9-week plan, export CSV
This post is how I built it
1. The real parent pain isn't "can't find camps." It's "can't organize them"
Google "2026 summer camp Taiwan." Hundreds of results
The problem isn't too little info. It's too much
Every organization has a different website, different signup dates, different age limits, prices ranging from $3,000 to $20,000 per week, some full-day, some half-day, some overnight
What a parent has to do:
- Collect the info
- Match against your kid's age and interests
- Map onto nine weeks
- Balance budget
- Sync with your partner
Average time: 10-20 hours per family. Every mom group I know does this dance every spring
Engineer instinct kicks in: this should be automated
2. Why I built it instead of using existing tools
There are parent apps. I looked at a few and decided to build my own
Three reasons:
- Incomplete data — most platforms only list partner organizations. The big names like NTNU and Godot Theatre Company aren't on them
- No nine-week planning — they're course catalogs. But what parents actually decide is "how do I plan the whole summer"
- Not customizable — my daughter's interests, my budget, my pickup logistics are all unique
So I built one. And turned it into an AI tool demo while I was at it
3. How I did it (it wasn't hard)
Honestly, I barely wrote code
Most of the time I had Claude Code open in terminal, telling AI what I wanted, and it built it
Rough flow:
Step 1: tell AI what I'm doing
I didn't hand it a schema or a tech spec. I just talked:
"I want a tool to help parents plan summer camps. The data is summer camps across Taiwan. I'll give you a list of organizations, you scrape 2026 info from each official site. I want a quiz, filtering, a draggable nine-week planner, CSV export. No DB, no server, just front-end localStorage"
That's it. Done
Step 2: let it scrape
I dumped the org list from mom group chats, Google searches, friend recommendations, all into Claude. Told it to scrape each official site
It got stuck, scraped wrong stuff, mixed up 2025 with 2026. Every couple hours I checked in. When something looked off I'd say "this org hasn't announced yet, mark it tbd, don't make stuff up"
Two days. 47 camps cataloged. Then a friend asked "what about Chiayi, Pingtung, the islands?" Another round. Final: 69 camps, 35 organizations, 22 cities
Step 3: UI/UX iteration
Rest of the time was looking at the screen. Is the quiz flow smooth, does filtering feel right, does the 9-week planner drag well
Every ugly thing got "this color should be cream, the button is red and feels dangerous, planner should collapse to the right side" and Claude fixed it
For the whole week my job was: say what I want → look at result → complain about ugly parts → it fixes → look again
I didn't deeply understand the schema, what TypeScript is, how the data structure should be designed. AI thinks about that. You just need to know what you want
Any dad in any town can build this. Really
4. Three things I learned
1. AI isn't "one-click magic." It's "the assistant that saves 70% of your time"
I probably hit Claude 200 times across the week. Each time a small problem: "this org changed URL," "W3 dates are wrong," "the quiz logic should..."
A parent organizing 47 camps by hand: 20 hours. Me building the whole tool, code included: 15-20 hours
But once the tool exists, other parents spend 0 hours using it
That's the value of tools. You build once. Everyone benefits
2. AI's superpower isn't writing code. It's "structuring unstructured data"
Turn web pages, PDFs, random LINE messages into a unified schema — this used to take an intern two weeks. Now Claude does it in two days
I helped a traditional org with digital transformation last year. The biggest bottleneck was exactly this. Wrote about it here
3. Parental decision cost is severely underrated
The hardest part of building this wasn't tech. It was figuring out "what does a parent weigh when planning nine weeks?"
I made a list:
- Easy pickup (location)
- Are friends going (social)
- How does it sync with partner's leave (timing)
- Budget (money)
- What does the kid want (interest)
- Don't burn out (alternate full-day and half-day weeks)
Each of these requires thinking. So the tool's job is to compress the info, return the choice to the parent — not decide for them
5. Who can use this
Good for:
- Kids entering kindergarten through 7th grade
- Greater Taipei area (some camps cover other cities)
- Parents who want to see all options at once instead of hopping across 30 official sites
Prices follow each official site. Only public info collected
6. If you want to do something similar
People keep asking me "AI is so powerful now, how do I start using it?"
My answer is always: find one thing you do every year, find it annoying, but it has a pattern. Have AI do it once
Not a big system. One specific small thing
For me it was friends always re-doing summer camp planning. So that became my project. Yours might be:
- Yearly travel planning
- Comparing insurance plans
- Tracking school notices for your kid
- Organizing elderly parents' medical records
Start small. Build it. Share it with the people around you who need it
That's what AI in practice actually looks like
If you want to know how I build products with AI, or want a custom version, book a call
If this tool helps you, please share it with other parents
My daughter enters 3rd grade this year. Next year I'll do this again. Hopefully by then there are better tools out there so I don't have to