SOKI（想起）mcp.kitemir.jp

Privacy Policy

Last updated: 24 October 2026

This English text is provided for convenience. The Japanese version is the original and the only legally binding one; if the two differ, the Japanese version prevails.

Introduction

SOKI holds the "memories" of the person using it and supplies them to the AI services that person has authorised. Given the nature of what is handled here, we treat it as important to state clearly what we do and do not collect.

1. Information we collect

· Account information (email address, password, display name, Google account identifier)

· Observations (statements of fact about you), recorded by you or by an AI during a conversation

· The utterance a record came from (the raw log)

· A record of which memories were read, and when (the recall log)

· The conversational stance you configured, dates for plans, and notification preferences

· Payment information (Stripe customer ID, invoice ID, amount, payment date)

· Access logs such as sign-in times, IP address and browser type, and cookies

· How often you use the "Notes" and "Growth record" features (a monthly count and the time of last use). We do not store the text of your notes, your requests to AI or its answers, or the content of growth records (see 6).

· In the "Self-introduction video" feature, a finished video you choose to "keep in SOKI" (it contains your face and voice and the text on screen), with its title, size and the time it was kept (see 7)

· A self-introduction video script sent from your AI (held temporarily until you load it on screen; see 7)

· A record of backgrounds made with AI for self-introduction videos (request number, time, where the image is held, and the count for the day; see 7). We do not store your documents, the videos you record, or the voice used for subtitles (see 7).

2. Information we do not collect

The following is not stored at all, by design.

· Judgements about a person, or assessments of character. An interpretation such as "meticulous" is not recorded; only the statement of fact behind it is.

· Credit card numbers. Payment is processed by Stripe and we never hold the card number.

Sensitive personal information (health, medical history, beliefs, social status, criminal record and the like) is not stored by default. Statements suspected of falling into this category are detected automatically and refused. Only if you give your consent on the settings screen can appointments such as a hospital or dental visit be stored (Article 20(2) of the Act on the Protection of Personal Information). You may withdraw that consent at any time, and anything stored can be reviewed and deleted by you from the dashboard.

Whether or not you consent, sensitive personal information is never put in the morning email (it is not even counted there) and is never used as material for a growth record.

The automatic detection is not perfect. If something is recorded by mistake, you can delete it yourself from the dashboard.

3. Information about people other than you

The service may handle statements about third parties such as family or friends (for example, "A's birthday is 6 August").

Third parties are recorded under pseudonyms only. The mapping between a real name and a pseudonym is not stored on our servers.

As a result, no specific individual can be identified from the information we hold.

If you type a real name instead of a pseudonym, that is what will be stored. Please use pseudonyms when entering information about other people.

4. Purposes of use

· Providing the service (holding memories, supplying them to AI services, surfacing items whose date is near, helping you search and tidy the notes on your computer, making growth records, and helping you make self-introduction videos: writing the script, making subtitles and keeping finished videos)

· Verifying identity and authenticating you

· Billing and payment

· Preventing misuse and investigating faults

· Responding to enquiries

· Complying with the law

5. Disclosure to external services

To the extent needed to run the service, information is provided to the following.

· AI services you have authorised (Claude, ChatGPT and so on): memories and your conversational stance are supplied. Which services receive them is limited to what you yourself allowed on the connection screen.

· Stripe (Ireland / United States): to process payments.

· Google (United States): to authenticate you if you sign in with a Google account.

· OpenAI (United States): when the memory search feature is used, the text of an observation is sent to obtain a numeric representation (an embedding). In addition, when you choose to send to AI in the "Notes" or "Growth record" features, the information described in 6 is sent.

· Anthropic (United States): when you choose to in the script step of the "Self-introduction video" feature, the information described in 7 is sent. Depending on our settings, OpenAI (United States) may be used instead of Anthropic.

· Google (United States): for subtitles in the "Self-introduction video" feature, the information described in 7 is sent to the Gemini API.

· fal (United States): when a background image is made with AI for a self-introduction video, a sentence describing a room is sent. No video, voice, documents or other personal information of yours is sent.

· Our hosting provider (in Japan): the service's data is held on servers in Japan.

· Where disclosure is required by law.

We do not provide your memories to any third party other than those above. They are not used for advertising, and they are not provided as training data.

6. The "Notes" and "Growth record" features

The "Notes" and "Growth record" features of the Service (used in a browser on a computer) handle information as follows.

Reading and writing the notes on your computer

· Notes are read from the folder you choose on your computer, inside your browser. The text of your notes is not sent to or stored on our servers.

· Writing to your notes (adding, fixing, moving files), the record of those changes and the previous versions are stored only inside the folder you chose. Anything is written only after you check the change on screen and choose to write it.

· So that you can use the same folder next time, the information needed to open it (not the text) is stored inside your browser.

Sending to AI

· Only when you choose "Ask or request AI" or "Make a growth record", the following is sent through our servers to OpenAI (United States). Before sending, you can check how much will be sent and what it is on screen.

　– your request, and the excerpts and key points chosen from your notes

　– the names of your note folders, and a writing example (file names and the settings at the top of a note)

　– for a growth record: the observations in the chosen period and the history of how they were replaced. Observations treated as sensitive personal information, and observations about people other than you, are not sent.

· Excerpts from your notes contain what you yourself wrote, as written (this may include things about your health or family). Please check the content on the confirmation screen before sending.

· Our servers only relay this information and do not store it (no text is left in the database, files or logs). The only thing we store is how often you use the features (a monthly count and the time of last use).

· AI's answers and the growth records you make are not stored on our servers either. A growth record is kept only by saving it to your computer (the folder you chose) or downloading it.

How OpenAI handles the data

· OpenAI states that it does not use data received through its API to train its models unless the customer explicitly opts in. We have not opted in.

· OpenAI states that it may retain API inputs and outputs for up to 30 days to provide the service and to identify abuse.

· When sending, we specify that responses are not to be stored on OpenAI's side (store: false).

7. The "Self-introduction video" feature

The "Self-introduction video" feature of the Service (used in a browser on a computer) handles information as follows.

Recording and finishing

· Your computer's camera and microphone are used only when you start recording.

· The video and voice you record are finished inside your browser (cutting out the person, adjusting brightness, shortening pauses, adding text and exporting) and saved to the folder you choose on your computer. The video you record is not sent to our servers (except the voice for "Subtitles" and "Keeping finished videos" below).

· So that you can use the same folder next time, the information needed to open it (not the video) and the background images made with AI are stored inside your browser.

Writing the script

· Only when you choose documents (CV, career history, entry sheet and the like) and run the reading step, the content of the documents is sent through our servers to Anthropic (United States) to pick out the facts needed for a script. Then the facts picked out and your answers to the questions are sent to write the script. Depending on our settings, OpenAI (United States) may be used instead of Anthropic.

· The documents contain your own information as written, such as your name, address, contact details, education and work history. You choose which documents are sent. The AI is instructed not to put phone numbers, email addresses, dates of birth, ages, street addresses and the like into the script.

· Our servers only relay the documents, the facts picked out and the script, and do not store them (no text is left in the database, files or logs).

Scripts sent from your AI

· If you receive a script from an AI service you have authorised, the script (name, city or ward of residence, headline, catchphrase and body) is held temporarily on our servers until you load it on screen. It is deleted when you load it, or after 24 hours.

Subtitles

· Only when subtitles are made, the voice (one minute at a time) and the script are sent through our servers to Google's Gemini API (United States) to be transcribed. The picture is not sent.

· Our servers only relay the voice and the script and do not store them.

Making a background image with AI

· When you choose to, a sentence describing a room is sent from our servers to fal (United States) to make an image. No video, voice, documents or other personal information of yours is sent.

· To count uses per day, we store the request number, time and where the image is held (the latest 10) and the count for the day. The image itself is not stored on our servers.

Keeping finished videos

· Only when you choose "keep in SOKI", the finished video is stored on our servers (in Japan). The video contains your face and voice and the text on screen (such as your name and the city or ward you live in).

· Up to three videos per person are kept. When a fourth is kept, the oldest is deleted. Videos are deleted automatically six months after they were kept. You can delete them at any time.

· Only you, signed in, can view or save the videos you kept.

How the AI providers handle the data

· Anthropic's commercial terms state that it does not train models on data received through its API. Anthropic also states that it deletes API inputs and outputs within 30 days as a rule (they may be kept longer to enforce its usage policy or where the law requires).

· Google states that it does not use data received through the paid Gemini API to improve its products, and that it logs inputs and outputs for a limited period to detect and prevent abuse. We use the paid Gemini API.

· How OpenAI handles data is described in 6.

Age

· The "Self-introduction video" feature is for people aged 18 or over, because the terms of the Gemini API, used for subtitles, do not allow its use in services used by people under 18.

8. Transfers of data across borders

Data is held on servers in Japan.

However, when you use an AI service you have authorised, payment (Stripe), Google sign-in, the memory search feature, or send to AI in the "Notes", "Growth record" or "Self-introduction video" features, information is transferred to the country in which that provider operates (the United States and others).

By using the service you consent to these transfers outside Japan.

About sending to OpenAI (information provided under Article 28 of Japan's Act on the Protection of Personal Information)

· Destination country: the United States of America (OpenAI is located in California)

· That country's system for protecting personal information: there is no comprehensive federal privacy law; there are sector-specific laws and state laws such as the California Consumer Privacy Act (CCPA). For details, see the information published by Japan's Personal Information Protection Commission (https://www.ppc.go.jp/enforcement/infoprovision/laws/offshore_report_america/).

· Measures taken by the recipient: OpenAI states that it does not use API data for training without explicit opt-in, that retention is up to 30 days, that data is encrypted at rest and in transit (AES-256, TLS 1.2 or later), and that it is audited by a third party (SOC 2 Type 2).

· In the "Notes" and "Growth record" features, anything is sent only after you check the content on screen and choose to send it.

About sending to Anthropic and Google (the "Self-introduction video" feature; information provided under Article 28 of Japan's Act on the Protection of Personal Information)

· Destination country: the United States of America (Anthropic and Google are both located in California)

· That country's system for protecting personal information: as in the OpenAI item above (see the information published by the Personal Information Protection Commission).

· Measures taken by the recipients: Anthropic states that it does not use API data for training, that it deletes inputs and outputs within 30 days as a rule, and that it has a data processing agreement (DPA). Google states that it does not use data from the paid Gemini API to improve its products, that logging to detect abuse is kept to a limited period, and that it processes data under its data processing agreement (DPA).

· In the "Self-introduction video" feature, anything is sent only when you run the reading step or make subtitles.

9. Cookies and authentication

Cookies and sessions are used to keep you signed in. You may disable them in your browser, but you will then be unable to sign in.

Connections to AI services use OAuth 2.1 and issue an access token. You can revoke a token at any time from the dashboard.

10. Security measures

· HTTPS encryption for all traffic

· Passwords stored hashed (never in plain text)

· Access tokens stored hashed, with expiry times

· A limit on sign-in attempts

· Automatic blocking of judgements about people, and of sensitive personal information by default (stored only where you have given consent)

· In the "Notes" and "Growth record" features, a design that leaves no note text or AI answers on our servers (relay only)

· In the "Self-introduction video" feature, a design that leaves no documents, scripts or voice on our servers (relay only), and in which recorded videos are finished inside your browser

· Kept self-introduction videos are stored where they cannot be read directly from outside, and are handed only to you, signed in

11. Retention

· If you close your account, use stops immediately and the related information is deleted after six months. During those six months it can be restored on request.

· After the free period, an account with no sign-in for six months or more also becomes subject to deletion.

· Deletion covers account information, observations, raw logs, recall logs, conversational stance and access tokens.

· Transaction records we are required by law to keep (payment amount, date, transaction ID) are retained for the necessary period in a form that does not identify an individual.

· The record of how often you use the "Notes" and "Growth record" features is also deleted together with your account.

· Kept self-introduction videos are deleted six months after they were kept (the same applies if you close your account). A script sent from your AI is deleted when you load it, or after 24 hours. The record of backgrounds made with AI is deleted together with your account.

12. Access, correction and deletion

From the dashboard you can, at any time:

· export all of the memories held about you

· delete an individual memory

· delete your account

· Note that your notes and growth records are on your own computer and we do not hold them (we hold only the usage count, which is included in the export).

· You can view, save or delete the self-introduction videos you kept at any time from the "Self-introduction video" screen. Recorded videos and finished videos (other than those you kept) are on your own computer and we do not hold them.

For any other request to access or correct information, please contact us. We will verify your identity and respond so far as is reasonable.

13. Use by minors

A person under 16 needs the consent of a legal guardian to use the service.

The "Self-introduction video" feature is for people aged 18 or over (see 7).

14. Changes

This policy may be revised as needed. Where a change is significant, we will announce it within the service.

15. Contact

For questions about this policy, please use the contact form or the email address above.

Operator information