クリエイティブ・創作
You Don't Need to Write Code to Build an App — Two Months That Changed Everything for a 37-Year-Old Office Administrator
機械翻訳 / Machine-translated
❤️ 0
🔖 0
⚡ 0
機械翻訳 / Machine-translated
Ten years ago, Sawai (37) closed her programming beginner's book for good. Variables, functions, object-oriented programming — none of it stuck, and she concluded that coding simply "wasn't for her."
Yet one dream never faded: to build something with her own hands that would genuinely help people.
Working in the general affairs department, every time she tried to streamline workflows with Excel or Google Sheets, she'd feel the pang of "if only I could build something more freely." But she didn't have the time or the resolve to become an engineer. With a husband and a third-grade daughter at home, attending evening classes was simply not realistic.
The turning point came in January 2026, with a single post that appeared in her social media feed.
"Don't write code — let AI write it for you. That's vibe coding."
She clicked the link with skepticism. The article explained that by using a conversational AI development tool, you could direct an app using plain, natural language and the AI would generate the code for you. "Create a login screen." "Add this field." — everyday phrases that become working programs.
That same evening, Sawai downloaded a conversational AI development tool that offered a free tier. She opened the interface and typed: "I want to build a simple equipment-management app."
Seconds later, code appeared, and a small form materialized on her screen.
"It's actually working."
That alone was enough to make her heart leap.
Her goal was clear: make managing office supplies simpler.
At the time, the status of borrowed equipment was tracked using a paper ledger and Excel, and figuring out who had what took far too long. She had long thought, "If only I could change this."
Sawai set aside 10:00 to 10:30 p.m. every night — after her daughter had gone to sleep — as her development time. She gave instructions to the AI tool, reviewed the generated code, and when something didn't work, she'd type "this part is throwing an error." The AI would then suggest a fix.
The first week was a string of errors. Screens would go completely blank; buttons would do nothing when pressed. Even so, as she continued going back and forth with the AI, patterns started to emerge — "this error means a required field is missing," "this error is a database connection problem."
By week three, the core feature — registering an item's name and the name of the person borrowing it — was complete. In week four she added a return-date alert, and in week five she refined the design to be more mobile-friendly.
Then, in early March, two months in, Sawai had three colleagues try the app.
"This is incredibly useful."
Those words from a colleague made every hour of effort worthwhile.
Equipment that had previously required hunting through a paper ledger could now be tracked at a glance on a smartphone. A notification arrived the day before each return deadline. The time needed to log a single borrow or return shrank from roughly three minutes to thirty seconds.
"I actually built an app."
That realization was the view from the other side of the beginner's book she had closed ten years ago.
What Sawai gained was not a programming language. It was the ability to have a conversation with AI.
"When I press this button, I want it to navigate to this screen."
"I'm getting an error — where did I go wrong?"
"I want the design to feel simpler."
The skill is in conveying thoughts like these in a way the AI can understand. It feels less like programming and more like delegating work to a junior colleague — you don't need to give perfect instructions; you shape the result together through trial and error.
Behind a conversational AI tool runs a language model with massive document-processing capabilities. As of April 2026, leading models can process more than 15,000 pages of text in a single pass and build on previous conversation history to offer the next suggestion.
This is a fundamentally different experience from "learning to program" as it used to be understood. There's no need to memorize syntax by rote; even if you don't fully understand what an error message means, the AI will translate it for you.
The equipment-management app is still in use at the office. Around 80 lending records are logged each month, and the number of unreturned items has been cut in half.
Sawai's next goal is a simple booking system for clients — she wants to make reserving the small meeting rooms her company manages far smoother. She's aiming to finish it by summer, stacking up those nightly thirty-minute sessions once again.
"I never imagined it could be this much fun to see something you want to build actually take shape in your own hands."
In her words lives a view that was invisible to her ten years ago.
Many people have given up on programming. But in 2026, that door is open again.
What you need is not perfect code — it's the desire to build something. A conversational AI becomes the partner that turns that desire into reality.
If there's something you once wanted to build but almost gave up on, try spending just thirty minutes tonight talking with an AI. It starts with putting your idea into words: "I want to make an app like this."
HAIIA's certification program offers a structured curriculum covering AI literacy from the basics through hands-on practice. HAIIA's three pillars explore in depth what self-realization through AI can look like. And if the thought of starting alone feels daunting, the community page lets you connect with others who are taking on the same challenge.
Turn your "someday" into "today" — starting now. Take that first step, together with AI.
Yes, you can. Even if, like Sawai, you gave up once before, that's perfectly fine. With vibe coding, code is generated simply by giving the AI instructions in natural language. What matters is having a clear idea of what you want to build; you don't need to memorize any syntax or grammar. Start with a small feature, keep up the conversation with the AI, and you'll gradually find yourself able to tackle more complex functionality.
It varies from person to person, but with thirty minutes a day, two to three months is a realistic timeframe for a basic business app. In Sawai's case, she dedicated thirty minutes every night starting at 10:00 p.m. and completed her equipment-management app in two months. The first week tends to be slow because of error-handling, but by week three the core functionality starts working, and by week five you can move on to design refinements. The key to success is building steadily, without rushing.
Many tools let you get started with a free tier. A free plan is more than enough at first; monthly fees only become relevant once you need more advanced features or need to generate large volumes of code. Most conversational AI development tools offer plans with sufficient capabilities for around 2,000–3,000 yen per month. Sawai started entirely for free during her first month, then switched to a paid monthly plan in her second month.
This article is reprinted from HAIIA Notes.