Healthcare’s information work,supported on one platform.

SERVICES

Four AI services on one platform

WORK

AI helps with four kinds of work

Choose the task that fits your work. In four steps, we show which service a conversation or record passes through, what it becomes, and who reviews and edits it.

1Choose a task

Documenting visitsConversation → draft noteM-RecPreparing documentsRecords → draft documentM-AutomateDeployedLooking things upQuestion → answer with sourcesM-ChatRecording meetingsDiscussion → draft minutesM-ConComing soon

Documenting visits: M-Rec drafts a clinical note from the conversation during the visit. The physician reviews and edits it, then pastes it into the electronic health record (EHR).

Preparing documents: From a month of nursing records, M-Automate creates a draft home-visit nursing report (the station’s monthly report to the attending physician) and a PDF. A nurse reviews and edits the report before it is sent.

Looking things up: M-Chat answers from the in-house documents you register and cites its sources. If the documents don’t cover a question, it is designed to answer “No match” rather than fill gaps with guesses.

Recording meetings: We are preparing a service that sorts what is said in meetings into four sections and drafts the minutes.

Watch a video of the actual screen on the M-Rec pageSee all 19 supported tasks on the M-Automate pageSee Clinical mode and General mode on the M-Chat pageSee the planned features on the M-Con page

2Click a number to show each stage.

Documenting visits (M-Rec)

Documenting visits: steps

1. 1Record the conversation
2. 2Identify key information
3. 3Draft note
4. 4Physician reviews and edits

M-Rec (AI)Physician

1Works in the clinic or on home visits (scheduled physician visits to the patient’s home). One tap records the conversation, and transcription runs as you record.

2Picks out the information needed for care from the conversation. The letter beside each highlight shows which section of the draft it goes into.

3Organizes it into a draft in the documentation style of each specialty and physician. It is designed not to add anything that wasn’t said.

4The physician reviews and edits the draft against the transcript, then pastes it into the EHR.

Recording and transcript

Home visit · Patient A (fictional), 84, femalePrompt: Home visitRecordingRecorded

“Hello. Any changes over the past two weeks?”

“How has she been at night, and how is she doing with her medications?”

“She gets up at night and searches through her dresser two or three times a weekS. We keep track of her medications on a calendar, and she hasn’t missed any dosesS.”

“How’s your back pain?”

“About the sameS.”

“Blood pressure is 132 over 76, pulse 72, oxygen 97%O. Nice and stableA.”

“You’re getting around the house holding onto furnitureO. Your forgetfulness is progressing gradually, but there’s been no sudden changeA.”

“Let’s be careful to avoid fallsP. I’ll talk with your care manager about increasing day service to three times a weekP.”

M-RecIdentifyOrganizeDraftSOAP

Clinical noteTranscript

(Before drafting)

SPer daughter, gets up at night to search for things 2–3 times/week. Medications managed with a calendar; no missed doses. Per patient, low back pain unchanged.

OBP 132/76, P 72, SpO2 97%. Walks holding onto furniture indoors.

AForgetfulness slowly progressing; no sudden change. BP stable.

PFall precautions. Discuss with care manager increasing day service adult day care to 3 times/week.

Edit requestAsk AI to editReplace “day service” with “adult day care”

Reviewed by physicianCopy all → paste into EHRTransfer buttonPlanned

Preparing documents (M-Automate)

Preparing documents: steps

1. 1Retrieve records
2. 2Identify key points
3. 3Draft report
4. 4Nurse reviews and edits

M-Automate (AI and robot)Nurse

1The robot pulls one month of nursing records from the EHR.

2AI reads the records and picks out changes in weight and symptoms, along with what was reported to the attending physician.

3AI drafts the “Clinical course” section. The robot files the report in the EHR as a draft and also creates a PDF.

4The nurse reviews and edits the draft, then sends it to the attending physician.

EHR · Nursing records · September 2026

Patient B (fictional), 81, male · Weekly visitsRobot: September records retrieved

9/1Weight 48.2 kg. Lower-leg edema (±).

9/8Weight 49.0 kg (+0.8 kg from last week). Lower-leg edema (+). Short of breath on stairsChange. Reported to attending physician; instructed to temporarily increase diureticResponse.

9/15Weight 48.4 kg. Discussed salt restriction with patient and daughterResponse.

9/29Weight 48.1 kg. No edema. No shortness of breathCurrent. No missed dosesCurrent.

M-AutomateIdentifyOrganizeDraftChangeResponseCurrent

Home-visit nursing report

To: Attending physician, Sakura Internal Medicine Clinic (fictional)

Fields the robot fills in

Client: Patient B (fictional) · Date: Sep 30, 2026 · Visits: Sep 1, 8, 15, 29

Clinical course

(Before drafting)

ChangeOn September 8, weight was up 0.8 kg from the previous week, with lower-leg edema and shortness of breath on stairs.

ResponseReported to the attending physician and was instructed to temporarily increase the diuretic. Discussed salt restriction with the patient and family.

CurrentAt the September 29 visit, no lower-leg edema or shortness of breath was observed. No missed medication doses. Next month, we will continue to monitor mainly for changes in weight and lower-leg edema.Added by nurse

Draft filed in EHRPDF: Report (Patient B, September)

Nurse reviews and editsSend to attending physician

Looking things up (M-Chat)

Looking things up: steps

1. 1Ask in everyday language
2. 2Search in-house documents
3. 3Answer with sources
4. 4Check the sources

M-Chat (AI)Staff

Ask something the documents don’t cover

1Ask what you want to know in everyday language. Questions that don’t include patient information use General mode.

2Searches the registered in-house documents for passages that answer the question.

3The answer is organized under headings and cites the sources it is based on.No relevant passage is found, so the answer is “No match.”

4Open the source and check the basis before using the answer.You can confirm right away that the registered in-house documents don’t cover this question.

M-Chat · Screen in development

Aoba Hospital (fictional)General mode

If a patient falls, who do we report to, and in what order?

How do I set up the new infusion pump?

Reference documents (registered)

Incident response manual3-2 Falls

Internal policies

In-house protocols

No relevant passage found

M-ChatSearchOrganizeAnswer3-2

Answer

(Before answering)

When a fall occurs

Make sure the patient is safe, and check for consciousness and injuries.

Reporting order

1 Attending physician (on-call physician at night)
2 Your unit’s nurse manager
3 Submit an incident report

Source: Incident response manual 3-2

No match: No description of this procedure was found in the registered in-house documents.

Incident response manual 3-2 Falls

When a fall occurs, ensure the patient’s safety and check for consciousness and injuries. Report to the attending physician (the on-call physician at night), then to your unit’s nurse manager, and submit an incident report.

Highlighting added for illustration.

Open the source to checkDesigned not to guess

Recording meetings (M-Con, coming soon)

Recording meetings: steps

1. 1Record the meeting
2. 2Identify key points
3. 3Draft minutes
4. 4Review, edit, and share

M-Con (AI)Meeting organizer

1Records the meeting live, or imports a recording of an online meeting or an audio file (planned).

2Transcribes it and picks out decisions, plus action items with owners and due dates (planned).

3Organizes the content into a draft with four sections: summary, decisions, action items, and agenda items (planned).

4We plan to let you review, edit, and save the minutes, and choose who at your facility can see them.

Meeting recordingComing soon

Aoba Hospital (fictional) · Pre-discharge conference · Patient C (fictional)RecordingRecorded

“Patient C is scheduled to be discharged next Tuesday.”

“There are no handrails in the bathroom at home, so we’re planning to have bathing done at day servicesDecision.”

“Home-visit nursing will start with weekly visitsDecision, and we’d like the nurses to check on medications.”

“The community liaison office will contact the home-visit nursing station by FridayAction item.”

“Let’s share another update about two weeks after dischargeAgenda item.”

M-ConIdentifyOrganizeDraftDecisionAction itemAgenda item

Draft minutes

(Before drafting)

Summary

Discussed support for bathing and medications at home as Patient C (fictional) prepares for discharge.

Decisions

DecisionHome-visit nursing to start with weekly visits. Plan to provide bathing through day services.

Action items

Action itemContact the home-visit nursing station (Owner: community liaison office / Due: Friday)

Agenda items

Agenda itemSupport for daily life after discharge; next update (about two weeks after discharge)

Review, edit, and saveSharing: everyone at the facility / creator only (planned)

Every task shares one platform, with the same sign-in, user management, and activity log. See the shared platform

How this diagram works

The diagram switches between content already written into the page, based on the task and step you choose. Even where scripts don’t run, the same content appears when you make a selection.

radio and :has()clip-pathcontainer queries@starting-style

In older browsers, all four tasks are shown in their finished state, stacked vertically and labeled by task.

*A sample to illustrate how it works. The medical institutions, patients, conversations, and records are all fictional. The M-Chat screen shows a version in development and may differ from the final product.

Talk to us about this task

ONE DAY

Each record feeds into the next task

Here is an example day at a hospital, a clinic, and a home-visit nursing station (an agency whose nurses care for patients at home). Choose each scene in turn to trace how records made earlier are used in later tasks.

1Choose a facility

HospitalClinicHome-visit nursing station

Aoba Hospital (fictional)Tue, Sep 8

2Choose a scene

9:10M-RecDocumenting visits11:30M-ChatReferral letter from notes14:00M-ConPre-discharge conference17:00M-AutomateSubmission list

Documenting visits: M-Rec drafts a clinical note from the visit conversation, using only what the recording picked up. The physician adds any findings not mentioned in the conversation.

Referral letter draft: M-Chat drafts a referral letter from this morning’s visit notes made in M-Rec. The physician reviews and edits the draft before using it.

Pre-discharge conference: We plan to create draft minutes from the meeting recording, organized into a summary, decisions, action items, and agenda items. M-Con is coming soon.

Submission list: M-Automate picks out items eligible for the outpatient data submission add-on from the day’s outpatient records and adds them to the list. Billing staff review and edit the list before submitting it.

8:009:0010:0011:0012:0013:0014:0015:0016:0017:0018:0019:00M-Rec M-Automate M-Chat M-ConComing soonReviewer9:1011:3014:0017:00Documenting visitsNotesPhysicianReferral letter draftReferralPhysicianSource recordM-Rec notesPre-discharge conferenceConfOrganizerSubmission listListBillingSource recordVia the EHR

Made by a serviceChecked by a personHow records connectExample of combined use

9:10

Internal medicine outpatient clinic

Physician

The physician sees Patient D (fictional), who is being treated for hypertension and reports shortness of breath. One tap starts recording the conversation.

M-RecDrafts a clinical note from the conversation

Recording and transcriptOutpatient internal medicineRecorded

Patient D (fictional)

“When did the shortness of breath start?”

“About a week agoS. Climbing stairs leaves me windedS.”

“Do you ever wake up at night coughing?”

“No, I don’tS.”

“Let’s get a chest X-ray and an ECGP right after this. For an echocardiogramP, I’ll refer you to cardiology. I’ll go over your blood test resultsP next week.”

Draft noteExcerpt

SShortness of breath on climbing stairs for about a week. No cough at night.

OBreath sounds clear, no heart murmur. No lower-leg edema.Added by physician

PChest X-ray and ECG done today. Referred to cardiology for echocardiography. Blood test results to be explained next week.

Reviewer

Physician9:20

Reviews and edits the draft against the transcript, and adds exam findings that weren’t part of the conversation.

Goes toPasted into the EHRTransfer button Planned

11:30

Internal medicine outpatient clinic

Physician

Between appointments, the physician prepares a cardiology referral for Patient D (fictional), seen this morning.

M-ChatDrafts a referral letter from the visit notes

M-ChatScreen in developmentClinical mode

Patient D (fictional)

Draft a referral letter to cardiology from today’s visit notes.

Reason for referral

I am referring this patient to you for an echocardiogram.

Symptom history

The patient has noticed shortness of breath when climbing stairs for about a week. No cough at night.

Tests today

A chest X-ray and ECG were performed. Blood test results are pending.

LinkM-Chat builds its draft from the visit notes created with M-Rec at 9:10.

Reviewer

Physician11:40

Reviews and edits the draft, adds the chest X-ray and ECG results, and finalizes the letter.

Goes toReferral letter to cardiology

14:00

Conference room

Physicians, nurses, community liaison staff, care managers, and others

The team holds a pre-discharge conference for Patient C (fictional), who is due to go home soon.

M-ConComing soonWill draft minutes from the meeting recording

Draft minutesPlanned feature

Summary

Discussed support for bathing and medications at home as Patient C (fictional) prepares for discharge.

Decisions

Home-visit nursing to start with weekly visits. Plan to provide bathing through day services.

Reviewer

Meeting organizer15:00

Reviews and edits the draft minutes, then saves them and shares them within the hospital (planned).

Goes toShared within the hospital

17:00

Billing department

Robot and AI → Billing staff

The list for the outpatient data submission add-on (a fee-schedule add-on for facilities that submit outpatient care data to Japan’s health ministry) is updated the same day.

M-AutomateDeployedPicks out eligible items and adds them to the list

17:00Robot retrieves17:02AI extracts17:05Robot adds to list18:00Staff review

Run frequency and times are set to fit each facility’s workflow.

Submission list: September

Sep 8Patient D (fictional)Internal medicineItems extracted

Sep 8Patient H (fictional)Internal medicineItems extracted

Sep 1–7112 entriesInternal medicine, etc.Reviewed

Counts are examples.

LinkM-Automate reads records in the EHR. The 9:10 outpatient notes are the ones the physician reviewed and edited before pasting them into the EHR. Example of combined use

At hospitals, other tasks we can handle include drafting discharge summaries and discharge support plans, and calculating length of stay by bed type.

Reviewer

Billing staff18:00

Reviews and edits the rows added that day, and checks the full list at month-end.

Goes toSubmitted for the outpatient data submission add-on

Sakura Internal Medicine Clinic (fictional)Tue, Sep 8

2Choose a scene

9:10M-RecOutpatient visit12:30M-ChatProgress summary14:00M-RecHome visit notes17:30M-AutomateHome care guidance report

Outpatient visit: From the conversation at a routine visit, M-Rec drafts a clinical note. The physician reviews and edits it, then pastes it into the EHR.

Progress summary: From notes on earlier visits created with M-Rec, M-Chat summarizes how the patient’s nights have been going. The physician checks the summary against the original notes.

Home visit notes: From the conversation recorded at the patient’s home, M-Rec drafts a clinical note. The physician reviews and edits it, then pastes it into the EHR.

Home care guidance report: The physician’s report to the patient’s care manager under long-term care insurance. M-Automate drafts it from records in the EHR. After the physician reviews and edits it, the robot sends it.

8:009:0010:0011:0012:0013:0014:0015:0016:0017:0018:0019:00M-Rec M-Automate M-Chat M-ConComing soonReviewerNot used in this example9:1012:3014:0017:30Outpatient visitClinicPhysicianProgress summaryRecapPhysicianHome visit notesHomePhysicianGuidance reportReportPhysicianSource recordVia the EHR

Made by a serviceChecked by a personHow records connectExample of combined use

9:10

Outpatient clinic

Physician

A routine visit for Patient E (fictional), who is being treated for hypertension and type 2 diabetes.

M-RecDrafts a clinical note from the conversation

Recording and transcriptOutpatientRecorded

Patient E (fictional)

“How has your blood pressure been at home?”

“In the mornings, the top number is around 130S. I’ve also lost 2 kilos over the past monthS.”

“How about snacking between meals?”

“I’ve given up my late-afternoon sweetsS.”

“Your blood pressure and weight are both heading the right wayA. Let’s keep your medication as it isP. We’ll do blood work next timeP.”

Draft noteExcerpt

SHome BP (morning): systolic around 130. Weight down 2 kg over the past month. Has stopped late-afternoon snacks.

ABP and weight both improving.

PContinue current prescriptions. Blood draw at next visit.

Reviewer

Physician9:20

Reviews and edits the draft against the transcript, then pastes it into the EHR.

Goes toEHR

12:30

Clinic

Physician

Before this afternoon’s home visit to Patient A (fictional), the physician reviews how things have gone at earlier visits.

M-ChatFinds earlier notes and summarizes progress

M-ChatScreen in developmentClinical mode

Summarize how nights have been going for Patient A (fictional).

Nighttime behavior over time

August 12 notes: Occasionally wakes during the night (per daughter).

August 25 notes: Gets up in the middle of the night to search for things about once a week (per daughter).

LinkM-Chat searches the notes created with M-Rec during the August visits.

Reviewer

Physician12:40

Checks the answer against the original notes and uses it to prepare for the afternoon visit.

Goes toPreparing for the afternoon home visit

14:00

Home of Patient A (fictional)

Physician

The physician records the conversation during the home visit on a smartphone.

M-RecDrafts a clinical note from the conversation

Recording and transcriptHome visitRecorded

Patient A (fictional), 84, femalePrompt: Home visit

“She gets up at night and searches through her dresser two or three times a weekS. Her medications are managed with a calendar, and she hasn’t missed any dosesS.”

“How’s your back pain?”

“About the sameS.”

“Blood pressure is 132 over 76, pulse 72, oxygen 97%O. Nice and stableA.”

“You’re getting around the house holding onto furnitureO. The forgetfulness is progressing little by little, but there’s no sudden changeA.”

Draft noteExcerpt

SPer daughter, gets up at night to search for things 2–3 times/week. Medications managed with a calendar; no missed doses. Per patient, low back pain unchanged.

OBP 132/76, P 72, SpO2 97%. Walks holding onto furniture indoors.

AForgetfulness slowly progressing; no sudden change. BP stable.

Records even where there’s no signal and keeps the recording on the device (currently up to one recording).

Reviewer

Physician16:00

Back at the clinic, reviews and edits the draft against the transcript, then pastes it into the EHR.

Goes toEHR

17:30

Clinic

Robot and AI → Physician → Robot

From the day’s visit notes, the clinic prepares a home care guidance report for the care manager who plans the patient’s long-term care services.

M-AutomateDeployedReads the notes and drafts the guidance report

17:30Robot retrieves17:32AI drafts18:10Physician reviews18:20Robot sends

Home care guidance report

Fields the robot fills in

Client: Patient A (fictional) / Visit date: Tue, Sep 8, 2026

Client’s condition and clinical course

Forgetfulness is progressing gradually, and she gets up at night to search for things 2–3 times a week. Her blood pressure is stable, and her low back pain has not worsened. To help prevent falls, please consider increasing adult day care to 3 times a week. If the nighttime behavior continues, we will consider reviewing her medication.Added by physician

Recipient: Kamome Care Management Office (fictional)

LinkM-Automate reads records in the EHR. The 14:00 visit notes are the ones the physician reviewed and edited before pasting them into the EHR. Example of combined use

Reviewer

Physician18:10

Reviews and edits the draft and adds the treatment plan. At 18:20, the robot turns it into a PDF and sends it.

Goes toKamome Care Management Office (fictional)

Minato Home-Visit Nursing Station (fictional)Wed, Sep 30, month-end

2Choose a scene

10:00M-AutomateHome-visit nursing report13:00M-AutomateFormatting notes16:00M-ChatUpdating the care team

Home-visit nursing report: From a month of nursing records, M-Automate drafts the report’s “Clinical course” section. The nurse reviews and edits it before it is sent.

Formatting notes: M-Automate tidies nursing notes entered by voice, including the measurements. After the nurse reviews and edits them, the robot enters them into the EHR.

Updating the care team: When the nurse pastes in today’s notes, M-Chat sorts out what to tell the attending physician and what to tell the care manager.

8:009:0010:0011:0012:0013:0014:0015:0016:0017:0018:0019:00M-Rec M-Automate M-Chat M-ConComing soonReviewerNot used in this exampleNot used in this example10:0013:0016:00Home-visit nursing reportReportNurseFormatting notesFormatNurseUpdating the care teamUpdateNurseSource recordPasted by the nurse

Made by a serviceChecked by a personHow records connectExample of combined use

10:00

Home-visit nursing station

Robot and AI → Nurse

It’s month-end. From September’s nursing records, the station prepares the home-visit nursing report for the attending physician.

M-AutomateDeployedReads a month of records and drafts the report

10:00Robot retrieves10:02AI drafts10:05Robot creates PDF11:30Nurse reviews

Home-visit nursing report

Fields the robot fills in

Client: Patient B (fictional) / Visit dates: Sep 1, 8, 15, 29

Clinical course

ChangeOn September 8, weight was up 0.8 kg from the previous week, with lower-leg edema and shortness of breath on stairs.

ResponseReported to the attending physician and was instructed to temporarily increase the diuretic. Discussed salt restriction with the patient and family.

CurrentAt the September 29 visit, no lower-leg edema or shortness of breath was observed. No missed medication doses. Next month, we will continue to monitor mainly for changes in weight and lower-leg edema.Added by nurse

To: Attending physician, Sakura Internal Medicine Clinic (fictional)

Reviewer

Nurse11:30

Reviews and edits the draft, adds the plan for next month, and sends the report to the attending physician.

Goes toAttending physician, Sakura Internal Medicine Clinic (fictional)

13:00

Home-visit nursing station

AI → Nurse → Robot

Notes dictated during the morning visits are tidied up so they’re easy to read.

M-AutomateDeployedOrganizes measurements and tidies the notes

Voice-entered nursing notesPatient F (fictional), Sep 30

“Um, blood pressure 132 over 78BP, sats 96SpO2, a little swelling in the legsEdema, and the patient says they feel better than yesterdayPatient.”

Tidied notesNursing record

SPatient: “Better than yesterday.”

OBP 132/78, SpO2 96%. Mild edema in the lower legs.

Reviewer

Nurse13:10

Reviews and edits the measurements and wording, then finalizes the note. The robot enters the finalized note into the EHR.

Goes toEHR (nursing records)

16:00

Home-visit nursing station

Nurse

The nurse updates the attending physician and the care manager on how Patient F (fictional) is doing today.

M-ChatSorts what to share by recipient

M-ChatScreen in developmentClinical mode

(Today’s notes pasted in) Sort out what to tell the attending physician and the care manager.

To the attending physician

BP 132/78, SpO2 96%. Mild lower-leg edema. The patient reports feeling “better than yesterday.”

To the care manager

At today’s visit, the swelling in the legs was mild. The patient told us they feel “better than yesterday.”

LinkThe nurse pastes the notes tidied at 13:00 into M-Chat.

Reviewer

Nurse16:10

Reviews and edits the message for each recipient before sending it.

Goes toAttending physician and care manager

How this diagram works

Choose a facility and a scene to see what each service creates, who reviews it, and when.

radio and :has()CSS Grid line numbersCSS variables and animationclip-pathcontainer queries

In older browsers, the 11 scenes are listed vertically by facility, without the timeline.

*Times and review points are set after we learn how your facility works.

PLATFORM

Hand off more tasks. Same entry point, same sign-in.

You can start with just M-Rec and add other services as you need them.

1Choose tasks to hand off (select one or more)

Documenting visitsM-RecPreparing documentsM-AutomateLooking things upM-ChatRecording meetingsM-ConComing soon

Your facilitySakura Internal Medicine Clinic (fictional)

PhysicianNurseAdmin staffRobot (RPA)EHR

MS-AIShared platform

Entry and verification

Entry for peopleSingle URL, MFA supported

Gateway for robotsClient certificate, IP restrictions

Staff URL1Unchanged

Permission checkChecked server-side

PhysicianM-RecM-Chat

NurseM-Chat

Admin staffM-Chat

RobotM-Automate

Set per user and per serviceActivity log: who, when, what actionAdmin screen: usage and feesIn development

User management1Unchanged

Services

M-RecClinical documentation support

M-AutomateWorkflow automation with AI and RPA

M-ChatConversational AI for medical institutions

M-ConMeeting minutesComing soonExpected to be available to add

If you choose both: M-Chat can search clinical records from M-Rec.

Services0123Expandable

Dedicated facility environmentCloud in Japan

Recordings and transcriptsRecords you createRegistered in-house documents

Dedicated environment1Unchanged

Nothing selected: Choose at least one task. The service for each task you choose joins the shared platform.

M-Rec only: The staff URL, sign-in, user management, and activity log are built into the platform from the start.

M-Rec and M-Chat: Adding a service doesn’t change the URL your staff use or how they sign in. It joins the shared platform without rebuilding your environment.

With M-Automate: Robots connect through a different gateway from people. User management and the activity log are shared with the other services.

With M-Con: M-Con is coming soon. If you already use other services, we expect it can be added without rebuilding your environment.

M-Chat only: Start with the service you need. The staff URL, sign-in, user management, and activity log are built into the platform from the start.

How this diagram works

The services you choose slot into the platform, and the path from your facility to its dedicated environment lights up step by step. The numbers on each level show what increases and what stays the same.

checkbox and :has()CSS animationtransition on translatecontainer queries

With animation stopped, or where scripts don’t run, the selected services and the numbers on each level still appear the same way.

*This schematic illustrates the setup and is not an actual screen. Before adoption, we’ll tell you which services run in your facility’s dedicated environment on cloud servers in Japan. How robots connect depends on your environment.

SAFETY

We define where information flows and where it stops

Where is information processed, who can access it, what is logged, and what is sent outside?
MS-AI sets these ground rules on its shared platform, not service by service.

1Choose a scene

1Record a visitPhysician · M-Rec2Request a referral letter draftPhysician · M-Chat3Collect outpatient recordsBilling department · M-Automate4Look up regulations on the webStaff · M-Chat5Search terms include patient informationStaff · M-Chat

In scenes with a round number, information passes through. In scenes with a square number, it stops partway.

Your facility Aoba Hospital (fictional)

Physician Staff Billing staff Robot (RPA) EHR

Scene 1: Request Physician: Record the outpatient visit with Patient D (fictional)

Scene 2: Request Physician: Draft a referral letter to cardiology from today’s visit notes

Scene 3: Retrieve Robot: Collect the day’s outpatient records from the EHR

Scene 4: Search Staff: home-visit nursing order validity period

Scene 5: Search Staff: (patient’s name) shortness of breath diuretic

Scene 1: Result Draft clinical note. The physician reviews and edits it, then pastes it into the EHR

Scene 2: Result Draft referral letter (reason for referral, course of symptoms, today’s tests). The physician reviews and edits it

Scene 3: Result Submission list. Billing staff review and edit it, then submit it

Scene 4: Answer Draws first on public agency pages, with links to sources

Scene 5: Notice Search canceled because it contained patient information

MS-AI Shared platform · Cloud in Japan

Entry and verification Separate entrances for people and robots

Entry for people Single URL, MFA supported

Gateway for robots Client certificate, IP restrictions

✓ Verified

Permission check Checked server-side

Scene 1: Physician → can use M-Rec

Scene 2: Physician → can use M-Chat

Scene 3: Robot → can use M-Automate

Scene 4: Staff → can use M-Chat

Scene 5: Staff → can use M-Chat

Set per user and per service

✓ Verified

Services

M-Rec Clinical documentation support

M-Automate Workflow automation with AI and RPA

M-Chat Conversational AI for medical institutions

M-Con Meeting minutes Coming soon

Scene 1: Conversation → transcript → draft clinical note

Scene 2: In Clinical mode, creates a draft from M-Rec clinical records

Scene 3: Extracts eligible items for the outpatient data submission add-on and builds the submission listDeployed

Scene 4: Search terms are checked before they’re sent

Scene 5: Search terms are checked before they’re sent

Pre-send check Looks for patient information

Scene 4: home-visit nursing order validity period

Scene 5: (patient’s name) shortness of breath diuretic

✓ No patient information

Patient information detected

Scene 5: Stops here

✓ Processed

Dedicated facility environment Stored encrypted

M-Rec clinical record Patient D (fictional), Sep 8 Recordings and transcripts

Scene 1: Recorded audio is deleted after transcription (backups after 7 days). The transcript and draft are stored

Scene 2: M-Chat refers to the clinical record from Scene 1

Scene 1: ✓ Stored

Scene 2: ✓ Referenced

Outside MS-AI (external services)

AI model training Not used Neither your inputs nor the generated output is used

Scene 4: home-visit nursing order validity period

Search terms sent (with the staff member’s own consent)

External search service Outside Japan

AI responses are checked too ✓

Scene 1: Recorded audio is deleted after transcription. Backups are also deleted automatically after 7 days.

Scene 2: Requests involving patient information are made in Clinical mode. In this mode, you can draft referral letters and other documents from M-Rec clinical records.

Scene 3: Client certificates and IP restrictions limit where the robot can connect from. Billing staff review and edit the list before submitting it.

Scene 4: The input is checked before sending. Search terms go to the external search service only if no patient information is found. AI responses are checked for patient information in the same way.

Scene 5: If patient information is detected, the search is canceled entirely. The search terms are not sent outside.

Activity log (sample) Who, when, what action

1. Sep 8, 9:18 Physician M-Rec · Draft created Signed
2. Sep 8, 11:30 Physician M-Chat · Request in Clinical mode Signed
3. Sep 8, 17:02 Robot M-Automate · Eligible items extracted Signed
4. Sep 9, 14:20 Staff M-Chat · Web search (passed check) Signed
5. Sep 9, 14:26 Staff M-Chat · Web search (canceled by check) Signed

Activity log entries are digitally signed.

How this diagram works

Step through each scene to see where information goes and where it stops.

radio and :has()CSS animationcustom propertiessubgridcontainer queries

With animation stopped, or where scripts don’t run, the highlighted path and the “Stops here” label appear just the same.

*This schematic illustrates the setup and is not an actual screen. Before adoption, we’ll tell you which services run in your facility’s dedicated environment on cloud servers in Japan. How robots connect depends on your environment.

Four things to check before you start

Our design follows the medical information security guidelines issued by Japan’s Ministry of Health, Labour and Welfare (MHLW), Ministry of Internal Affairs and Communications (MIC), and Ministry of Economy, Trade and Industry (METI). For each deployment, we set out which responsibilities rest with your facility and which rest with MedSync.

Where is it processed?

For applicable services, all processing, including AI, stays in Japan. Before adoption, we’ll tell you which services this covers. Data is stored in your facility’s dedicated environment in the cloud, with encryption in transit and at rest. MS-AI can also be deployed on a closed network that is not connected to the internet.

Who can access it?

We use multi-factor authentication (MFA) and set permissions per user and per service. The server checks which services each user can access. Sessions lock automatically 180 minutes after sign-in.

What is logged?

We log who did what, and when. The activity log does not include patient names or the text of queries sent to the AI. Each entry is digitally signed, so you can check later whether it has been tampered with.

What is sent outside?

When an external search service is used, the input is checked before it is sent. If patient information is detected, the search is canceled entirely. This search works only after the connection to the external service has been set up and each staff member who uses it has personally consented to search terms being sent to a search service outside Japan. Neither your inputs nor the generated output is used to train AI models.

*Detailed specifications vary by service and contract.

Talk to us about data handling

FAQ

Frequently asked questions

CONTACT

Start by telling us how you work today

We’ll learn how you work today, then map out which tasks AI and RPA can streamline and which steps need a person to review. You’re welcome to talk to us even if you haven’t decided which service to start with.