I Built a Smaller Control Center for Korean Study, Everyday Plans, and Writing

Some days I have a Korean phrase to revisit, a practical question about studying abroad, and an interesting AI project I want to investigate for an article. Collecting them is easy. Deciding what I can actually do first is harder. I wanted a place that would turn a pile of “important” notes into one manageable next step.

Hi, I’m mio. I’m Japanese, and I’m learning English and Korean while looking for practical ways to make studying and daily life easier with AI.

I found the idea in a builder’s much larger business dashboard, then made an independent, smaller tool for my own kind of day. This free article shows what I built, the real screens and tests, what it cannot do, and a paper version you can try without buying anything. The build instructions currently available are in the paid section of my Japanese note article, which begins with a free introduction; there is no English paid edition to purchase here yet.

On this page

• The business dashboard that made me rethink my own day
• What my smaller “Daily Desk” actually does
• Public RSS goes into a reading queue, not a truth machine
• The uses I can imagine—and the ones I have not lived yet
• The local limits I would want to know before relying on it
• A free three-box test before installing anything
• The build guide that exists today is in Japanese

The business dashboard that made me rethink my own day

Matt Wolfe’s video, “I Built a FREE App That Runs Your Entire Business,” introduced me to his Control Center. The accompanying public repository describes a local-first dashboard for industry updates, verified mentions, newsletter monitoring, public audience figures, reminders, and tasks. Its code is already available for free. I am not selling a translated copy of that repository or suggesting that my small app can do everything his does.

What stayed with me was the less glamorous part of the story. The account made me think about the distance between an initial dashboard and a tool I would actually rely on. A quick scaffold is not the same as a tool you trust when something goes wrong. That made me ask a more personal question: if I have study notes, life admin, and article ideas competing for attention, what is the smallest dashboard that helps me decide rather than merely display more information?

The video tells the builder’s story; the repository lets you inspect the public implementation. Both are Matt’s work, not evidence that I reproduced his business features or results.

What my smaller “Daily Desk” actually does

I built mio Daily Desk independently around three lanes: Study, Life, and Writing. A note can have a title, a concrete next action, a due date, an importance level, a status, and a source URL when there is one. That is more useful to me than a long undifferentiated list. A reminder to “work on Korean” is vague; “choose three expressions I could not catch and say them aloud” tells me where to begin.

The opening sample is fictional. It contains four example notes, one already complete. The Today view therefore shows three unfinished candidates, including a Korean expression exercise, a check of an AI builder’s primary source, and a study-abroad preparation item. The app orders them using fixed rules involving status, due date, and importance. AI is not secretly deciding what matters. I can disagree with the order and change the underlying note.

The screenshot uses the same Japanese interface file as the note article. Its heading means “Start small: today’s three”; the three rows are Study, Writing, and Life examples. The panel on the right is a short “note to self,” not an AI-generated prediction. If I want to discuss a task with an AI assistant, the app prepares a handoff text for me to review and copy manually. It does not send my notes to Codex or any other model by itself.

The distinction matters because the sample is a test of the interface, not a diary of a day I spent studying in Korea. I have not measured whether this changes my grades or saves a particular number of minutes.

Public RSS goes into a reading queue, not a truth machine

For article research, I also added a deliberately manual way to collect headlines from a public HTTPS RSS or Atom feed. I enter a feed URL and choose “refresh now.” The app records the items it can read and avoids adding the same normalized URL twice. If network retrieval fails, I can import an RSS/Atom XML file that I have permission to use instead.

In a September 28, 2026 test, a third-party RSS feed for Hacker News supplied 20 items. Refreshing it again added zero new items and identified 20 URL duplicates. That is a test of ingestion and URL-level deduplication, not a claim that the feed is operated by Hacker News, that those articles are accurate, or that two different URLs about the same event will be merged. A separate local XML sample added two items and held one back. The content still needs to be read at its original source.

The Japanese heading means “Find things, but do not accept them uncritically.” Below that are “Public RSS URL,” “Refresh now,” and a file chooser for when the network route is unavailable. The retrieved headlines appear as Writing-lane notes for later checking; they are not verified summaries. The server accepts only constrained public HTTPS feeds and rejects local or private network targets. Those safeguards reduce risk, but they do not make every online source reliable.

The uses I can imagine—and the ones I have not lived yet

Right now, the most natural uses for me are Korean study and writing. I can turn a vague intention into a small action, attach the original page to an article idea, and leave a question marked as unresolved instead of letting an AI fill it in. That is a way to organize my next check, not a system that validates facts for me.

If I study in Korea next year, public university notices or official application pages might become useful sources to track. That is a future use case, not a report from living there. Private class handouts, a friend’s messages, and a part-time employer’s rota require different permissions and safeguards; “the app can store a note” is not permission to feed someone else’s information to an AI service.

The same boundary applies to work later on. Public research notes may fit a personal desk, while customer records or internal documents would not belong in this version. I would rather make the first useful case small and explainable than promise one dashboard for every part of life.

The shared diagram is in Japanese, so here is its route in English: hand-entered Study/Life/Writing notes and manually obtained public RSS/XML enter a list in the local browser. A fixed priority rule helps form the Today view; a person checks the short brief and any source link. The final box explicitly says that AI is not connected automatically. I kept the Japanese image file unchanged across note and WordPress rather than passing a translated diagram off as the same finished asset.

The local limits I would want to know before relying on it

This version runs a small Node server bound to 127.0.0.1 on my computer and opens in my browser. The notes live in that browser’s localStorage, in plain text, for that local origin. It does not have an online account, cloud sync, automatic backup, or a shared workspace. A different browser profile or local port does not automatically see the same notes. I tested exporting a JSON backup, previewing its contents, replacing the local state, and reloading the restored six-note/one-feed sample; a backup still has to be made and protected by the user.

The app itself makes no AI API calls and needs no AI API key. It does not read Gmail or a private school site, publish an article, set reminders on another service, refresh feeds on a schedule, or share a live page with friends. Clicking manual RSS refresh does make a network request to the chosen public feed. Anyone who wants a hosted, multi-person service would need authentication, hosting, privacy work, and a new round of testing; I have not done that here.

On macOS with Node.js 26 and Chrome 153, the local app’s 19 automated tests passed. I checked screenshots at desktop, 390-pixel, and 320-pixel widths, plus a real export-and-restore flow. The narrow screenshot below is a browser viewport test, not proof that my phone can reach a server running on my computer. Node.js 22 is the stated intended minimum, but I have not tested that version or Windows on a separate machine. Keyboard and reduced-motion support have code-level checks, not a complete device-level accessibility audit.

A free three-box test before installing anything

You can see whether this idea helps without using my app or Matt’s. On paper or in a notes app, make three boxes: Study, Life, and Writing. Put just one open item in each. Under each item, write the next action you could actually perform, a due date only if it is real, and a public source URL if the task depends on a published fact.

For example, “learn more Korean” becomes “say aloud three expressions I missed yesterday.” “Write an AI article” becomes “open the maker’s demo and note what the input and output really are.” If all three still feel too large, make each first action smaller. This is the decision I wanted my screen to support; you may discover that a sheet of paper is enough.

This approach suits someone juggling personal study, research, or small projects and wanting source links nearby. It is a poor fit for a team calendar, private-message automation, or someone who needs a fully hosted app on a phone. If Matt’s full business dashboard is what you want, his original repository is the place to inspect first.

The build guide that exists today is in Japanese

I published the Daily Desk story and the tested starter kit together in one Japanese note article. Its paid section covers the independent ten-file kit, fictional sample data, local setup and tests, manual RSS and XML checks, backup and restore, and full prompts for asking Codex to rebuild or adapt the tool. The distributed kit passed its tests; the reader-facing prompts were reconstructed from the build process and were not independently proven to reproduce the same app from an empty folder in one pass. The kit does not contain my portrait, other article images, account names, API keys, or Matt’s code. The app can be run without Codex; using Codex to modify it is a separate choice with terms and costs to check.

That guide is paid and in Japanese. Its regular price, checked on September 29, 2026, is ¥980, and the article shows that refund requests are accepted; an eligible repost can display a promotional ¥100 purchase route. Running the kit needs a computer on which you can install Node.js and a modern browser. I have not released an English paid WordPress edition, so there is no English purchase button or promise of English-language build support here. The confirmed tests are local tests, not proof that a purchaser has completed checkout or downloaded the attachment. If you only need the concept, the three-box exercise above is free; if you read Japanese and want the files and the tested setup route, the published note article is the existing route.

Production note: The desk scenes featuring mio are generated images made with her fixed reference portrait; they are not documentary photographs of a study day. The application images are screenshots of the tested local version, using fictional sample notes. The flow diagram is an explanatory graphic. Screens, tests, and operating limits above refer to the September 28, 2026 local version; future study-abroad uses remain ideas to test.