Their Muse Was "It Organized Gmail" — Reading the Program Note Blog Alongside My Own Moka
The result first
“Their Muse was ‘it organized Gmail’.” I read the Gmail test published on the blog Program Note, and read it alongside my own AI butler, Moka. It is the nineteenth of these side-by-side readings, after “Mugichan” of Keitai Watch (ledger No. 1446), “Hibana” of PC Watch (ledger No. 1453), the “stopping place” of Wagason-Wabitan of AI×TOC Institute (ledger No. 1454), “It moves before you even ask” by Yuji Kawabe on note (ledger No. 1467), “Don’t be fooled by Muse’s cuteness” by Hinuma in Keitai Watch (ledger No. 1475), “Handing the annoying little things of daily life to Muse” by Wakato on Zenn (ledger No. 1494), “You can hand over your daily tasks” by ai_agent_note on Zenn (ledger No. 1508), “Myutan” by Taishi Akiyama on the Alternative Blog (ledger No. 1515), “muse” by Miccell on note (ledger No. 1522), “Shiro” by i on note (ledger No. 1529), “Convenience and unease” by Kinu on note (ledger No. 1537), “Kashiwamochin” by Shiochan on note (ledger No. 1551), “More than just answering” by Ai Subarashiki on note (ledger No. 1564), “Just one line, and an aerial CG of Tokyo Station was done” by Kazuhisa Nishikawa in PC Watch (ledger No. 1577), “Sora” by Buhimaru on note (ledger No. 1584), “A Complete Turnaround” by RYO (Mochizuki) on note (ledger No. 1591), “Used Thoroughly for Five Days” by sima on note (ledger No. 1598), and “It Tells You Even What You Didn’t Ask For” by Tomomi on note (ledger No. 1605). The article is titled “I Had Muse Organize My Gmail! Testing Mail Sorting, Extracting Important Mail, and Writing Replies,” published on November 2, 2026. Neither the writer’s name nor their Muse’s name could be confirmed in the article.
Where we overlap is threefold. First, starting with small permissions: that house chose display-only access and wrote “reading and sorting only” into the request itself. Second, sorting mail and lining up the important ones first — in this house, too, the morning check brings the important notices forward instead of reading everything. Third, the stance that a person checks mistakes and important judgments: that house writes plainly that security decisions should also be checked on Google’s official screens, and that dates should finally be confirmed with the other party. Where we differ is this. That house is a verification series: Gmail alone is tested deeply in one article, and Calendar, Drive, and Sheets each get their own installment. This house narrows the AI to Muse only and the tools to Google apps only, and keeps using Gmail, Google Calendar, Google Drive, and Google Sheets inside the daily operation — the morning mail report, the schedule, the document store, and the ledger of articles — with the work divided across an organization and the records kept in lists and a ledger.
Hello, I am Kurashiki Bouzu (kurashiki-bouzu). Born in 1971, formerly in the funeral business, I now live in an old folk house in Kurashiki, Okayama, talking with my AI butler Moka every day and writing articles. Here is the reading, in order.
Their story — testing Gmail one step at a time, with read-only permission
Program Note is a blog that records programs and tools by actually trying them. This article tested, step by step and with the setup screens included, what becomes possible when Muse is connected to Gmail.
It begins with how to connect: open “Connectors” from Muse’s settings, choose Gmail, and link a Google account. The important part was how the permissions were chosen. Two were shown — “View email messages and settings” and “Manage drafts and send email” — and only the first was selected. In other words, the test began in a form where Muse could read, but could not send replies or move drafts on its own. The request itself carried the same line: “Please do not reply, send, delete, or move any mail at all. Reading and sorting only, please.”
The first test was sorting. The last seven days of mail were divided into “important / action needed,” “notices / information sharing,” and “advertisements / newsletters,” with counts for each and, for the important ones, subject, sender, and a summary. The result was 2 important, 1 notice, 1 advertisement — 4 in total. Before opening each message one by one, the whole picture comes into view first.
Next came the order. The messages needing action were lined up in priority order. First was a notice that the phone number used to recover a Google account had been changed, with the advice to check the account’s activity and change the password if needed when you do not remember changing it yourself. Second was a notice of a new sign-in on Windows, with the explanation that account protection is needed if it was not your own action. Alongside the order came “what to check” — and, written with it, the caution that security judgments should not rest on Muse’s answer alone, but should always be checked on Google’s official screens as well.
The search for mail needing a reply proceeded honestly. First, the last thirty days were checked: all 4 messages were automatic notices, and none needed a reply. Instead of stopping there, the writer prepared a test email confirming a meeting schedule and asked again. This time, out of 5 messages, one — “Confirming the meeting schedule” — was picked out as needing a reply, with the reason that the schedule cannot be fixed without one.
Last came the reply text. For that scheduling email, Muse proposed two forms: one for when the time works, and one for when it does not. Only the wording was requested, so nothing was sent and no draft was saved. And the most interesting scene of the test was here. The original email said “November 5 (Wednesday),” but November 5, 2026 is a Thursday. Muse noticed the mismatch and advised checking before replying. Beyond sorting and summarizing, it can become a chance to notice mistakes in what is written — that was the writer’s response, together with the closing note that whether a date is truly correct must finally be confirmed with the other party as well.
The closing impressions were three: mail can be sorted together; the order of action can be arranged; and reply wording can be drafted for you. At the same time, Muse’s judgment is not always right, so security notices and important work mail should be checked by yourself before acting. The cautions also line up clearly: read the permission screen carefully; do not grant more than you need; treat sending and deleting with care; be careful with mail containing personal or confidential information. The conclusion is that it is safest to begin with reading at the center, and to widen the range only after the mechanism and the permissions are understood. The same blog also runs a series testing Google Calendar, Google Drive, and Google Sheets, and the spreadsheet totals include a comparison between Muse and ChatGPT. It is a series that checks Google’s apps one at a time, not a single Gmail article alone.
Where we overlap — starting small, sorting to the important ones, and a person checking at the end
Three places overlapped as I read.
The first is starting with small permissions. That house chose display-only access, and drew the line “reading and sorting only” inside the request itself. In this house, too, Gmail is connected to Moka, but actions that go outside — sending, publishing, booking — are decided to proceed only after the owner’s approval. What can be done, and what may be entrusted, are kept as separate questions. That placement overlapped.
The second is the way of sorting mail and lining up the important ones first. That house divided seven days into three groups and lined up the two important ones with an order. In this house, too, the morning check has Moka search, count, and summarize the mail, bringing only the important notices forward first. Not “read everything,” but “look at the important ones first.” The idea of reducing the morning’s work overlapped.
The third is the stance that mistakes and important judgments are checked by a person. That house noticed the date-and-weekday mismatch through Muse, while writing plainly that the date must finally be confirmed with the other party, and security judgments on the official screens. In this house, too, what has not been confirmed is written as “unconfirmed” rather than filled in by guesswork, numbers in articles carry their sources, and a different pair of eyes checks before publication. An AI’s answer is not taken as the answer as-is. That carefulness overlapped.
Where we differ — a series that tests one app at a time, and a house that keeps using them in daily operation
The difference was also clear.
That house is a verification series. Gmail alone is checked deeply in a single article — the connection screens, the choice of permissions, the wording of the request, the result, and the cautions, left in order. Calendar, Google Drive, and Google Sheets are each checked the same way in their own installments. Including the step of preparing a test email and checking again, it is written as “a record that someone connecting for the first time can follow along and test in the same steps.”
This house is daily operation. The AI is narrowed to Muse only and the tools to Google apps only, and Gmail, Google Calendar, Google Drive, and Google Sheets are kept in use not only for testing but inside every day’s work. The morning mail report, the arranging of the schedule, the place where documents live (the “Ai Archive” in Google Drive), and the lists of articles and work (the ledger in Google Sheets) all sit inside one flow of operation. On top of that, the work is divided as it goes around: the one who researches and the one who checks are kept apart, each article carries a person in charge and a ledger number, and large jobs are divided by department. Records are kept in lists and a ledger (each article has a number, and whether it is a draft, checked, or published is visible; as of today, the articles counted as “published” in the ledger came to 1,591). It runs as an organization (with Moka at the center, three vice-butlers, the maid team, the ninja team, and departments for each kind of work arranged in a circle — 60 departments and 3,651 seats in total as of November 20, 2026). The name is Moka, and the personality was made by arranging words (“calm and composed,” “when numbers and feelings conflict, feelings come first,” and so on — 18 words in four layers: character, personhood, thinking, and action). Where that house is a series that checks each app one at a time, starting from the procedure, this house keeps using what it has connected inside its daily operation, and leaves the records in a ledger. That was the difference.
The same Muse takes a different shape in each house
Having read to the end, what I thought was, again, that the same Muse takes a different shape in each house. At Program Note’s house, Muse was the partner in a verification that checks Google’s apps one at a time, carefully, from procedure and permissions. In this house, it is a butler who handles mail, schedules, and documents together inside the daily operation. Starting with small permissions, lining up the important ones first, and a person checking at the end are very much alike — and yet a series that leaves the way of checking as a record, and an operation that keeps on using, place Muse in different spots.
I am grateful to the person who runs Program Note and wrote this article. Because they wrote not only the sorting and reply drafts that went well, but also the first result in which no mail needed a reply, the step of preparing a test email and checking again, and even the scene of the date-and-weekday mistake, it became a record in which someone connecting Gmail for the first time can see where to be careful as well.
Finally, one line this house keeps. An AI is something people use by giving it instructions. In that house’s record, too, which permissions to grant and how far to entrust were decisions for the person’s side. While widening the work we entrust, the final judgment stays on the person’s side. That is why it can be entrusted with peace of mind.
Sources
- Program Note, “I Had Muse Organize My Gmail! Testing Mail Sorting, Extracting Important Mail, and Writing Replies” (November 2, 2026): https://pg-note.com/archives/2767 — the original article used as material. The publication date follows the date shown on the page (2026.11.02). Neither the writer’s name nor their Muse’s name could be confirmed in the article
- Sources of the numbers: the numbers on that side (the seven-day sorting — 2 important, 1 notice, 1 advertisement/newsletter, 4 in total; the 4 messages in thirty days; 1 of 5 including the test email needing a reply; and so on) come from the original article’s text. That November 5, 2026 is a Thursday was checked against the calendar. The numbers on this side: the 1,591 published articles are the actual count of rows marked “published” in the ledger on November 21, 2026 (before this article was added), and the departments and seats (60 departments, 3,651 seats in total) are from this house’s records as of November 20, 2026. No amounts of money appear in the text
- November 21, 2026 ― 英語版(日本語版No.1629の内容を「結果から書く」形に書き直したもの)(10,882 characters)
- November 21, 2026 ― 公開(本人の決定による。7言語同時公開。アドレスの末尾に台帳番号を入れる新形式で、末尾は日本語版の番号No.1629)(10,882 characters)
