For student wellness and counseling centers

Stepped care for every student: AI handles the routine, your team decides

Screening, stepped care, care teams and follow-up in one place, in your center's own terms. AI proxies answer routine questions and check-ins within limits the provider and the student set, and messages about self-harm always go to people.

  • Screening scores suggest a care level; a provider decides
  • Counselor, nurse, case manager, health coach and peer supporter in one care team
  • Students choose what their AI proxy may share
Book a demoContact us
A student's detail view with the suggested care level and the care team.
A student's record: suggested care level, current level and care team.

Three kinds of accounts, one platform

Everyone signs in at the same place and gets a dashboard for their part: the center admin sets the center up, the care team works with students, and students follow their plan.

Center admin

The group admin who sets up the center. It is a provider account with admin rights, so Dr. Park also counsels students.

  • Terms: what the center calls its students, providers and care team, in English and Chinese
  • Roles and permissions: seven provider roles to start with; rename them, add roles or change any permission
  • Stepped care: the care levels and the screening-score rules that suggest a level
  • Policies and members: how conversations are deleted, whether students may remove their provider, the AI document format; provider accounts, group admins and invite links with a role, expiry and use limit
  • EHR connection: link the center's electronic health record through its standard FHIR API (SMART on FHIR)

When one installation serves several centers, a site administrator can open the same settings for each of them.

Recording: Dr. Elena Park opens Group settings and goes through its tabs: Terms, the Roles & permissions matrix (scrolled to the EHR columns), Stepped care levels and suggestion rules, Policies, Members with invite links (link addresses blurred) and EHR connection.
▶ Live demoCenter admin: Group settings, tab by tab.

How it works

From the first screening to follow-up, organised as stepped care: the least intensive help that works first, more when it's needed.

  1. Step 1

    Intake and screening

    Assign intake and screening questionnaires such as PHQ-9, GAD-7 and the C-SSRS screener. Students complete them from their own dashboard, and the scores are calculated on our servers.

    • Students answer on their own dashboard
    • Results are kept with their date
    Recording: Maya opens her assigned PHQ-9 from her Today list, answers the nine questions, sees the computed score and submits; the survey is marked Completed with a score of 6.
    ▶ Live demoMaya answers her PHQ-9; the score is calculated and kept with the date.
  2. Step 2

    Stepped care: a suggestion, a human decision

    Your levels run from self-help to intensive care or referral. The latest PHQ-9, GAD-7 and C-SSRS scores suggest a level, and a positive C-SSRS screener is flagged as urgent. A provider with the right permission sets the level; the system never does.

    • Levels and score rules are yours to edit
    • Who set the level, and when, is recorded
    Recording: Dr. Park opens Maya's record and reads the care level suggested by her PHQ-9 and GAD-7 scores. She applies it, and the level now shows it was set by Dr. Elena Park, with the date. Then she adds Daniel Kim, case manager, to Maya's care team.
    ▶ Live demoThe counselor applies the suggested level, and who set it is recorded; then she adds a case manager to the team.
  3. Step 3

    A care team around each student

    A student can have several providers: a counselor as the primary, plus a nurse, a case manager, a health coach or a peer supporter. Each sees what their role allows; a peer supporter, for example, never sees clinical notes.

    • One primary provider; others join or leave the team
    • The student sees who is on their team
    Group settings, Members: the center's providers with their roles (case manager, counselor, nurse, peer supporter).
    The center's providers and roles; a student's care team is drawn from them.
  4. Step 4

    Follow-up between sessions

    Care plans turn into recurring check-ins: mood, sleep, a short journal, questionnaires. The student checks them off each day, and a one-click report shows the trends.

    • Repeat PHQ-9 or GAD-7 to measure progress
    • Printable report with charts
    Recording: Maya rates her mood and stress, adds a note and saves, then logs 7.5 hours of sleep and saves. Both tasks turn Done, and today's row in the 14-day history and the streaks update.
    ▶ Live demoMaya completes her check-in and sleep log; the history and streaks update.
  5. Step 5

    AI proxies in messages

    The counselor's proxy answers routine questions (hours, how to prepare, rescheduling) from material the counselor approved and hands the rest over as a question with a draft. When a provider asks how a student is doing, the student's proxy answers only from what the student chose to share, with dates, and asks the student when the data is missing or more than 7 days old.

    • Every AI message is labelled as AI
    • Answers cite their sources
    Recording: Maya asks whether counseling sessions cost anything. Dr. Park's AI proxy answers, labelled as an AI proxy, and cites the Wellness Center's services page. Then Dr. Park asks Maya's AI proxy how her sleep has been; it answers from the sleep log Maya shares, with dates, and cites it.
    ▶ Live demoBoth proxies at work: every answer is labelled as AI and shows its sources and dates.
  6. Step 6

    The student decides what is shared

    A student who turns on their proxy chooses what it may draw on: check-ins from chosen care plans, their own answers, assessment scores. Anything not shared stays out of its answers.

    • Off until the student turns it on
    • Sharing can be changed at any time
    Recording: Maya opens her AI proxy settings, starts sharing her Mindful breathing plan, stops sharing her assessment scores and saves; “Sharing saved” appears.
    ▶ Live demoMaya changes what her proxy may share, and saves.
  7. Step 7

    Research and reports with the AI agent

    Staff can hand multi-step work to the AI agent: finding recent evidence, reading documents and web pages, and drafting a handout or an HTML report with its sources. The task keeps running while the counselor works elsewhere in the dashboard; running tasks stay at the top of the history, which can be searched and sorted. What the agent writes is a draft: the counselor checks the sources and the content before using it.

    • Every step is shown: searches, pages read, files written
    • A draft for staff to review, not clinical advice
    Recording: Dr. Park opens the AI Agent tab, clicks New task and picks General task, then types: “Find recent evidence on brief CBT-based interventions for exam stress in university students and write an HTML report with sources”. The agent starts searching the web and reading pages. She goes back to the history, where the task is pinned on top as Running with its elapsed time. About four minutes later it is completed; she opens it, widens the report and scrolls from its title and executive summary to its list of key sources. Back in the history, she searches “exam” and finds the task, then sorts the list by oldest first.
    ▶ Live demoA counselor hands off an evidence search; the agent searches, reads and writes an HTML report with its sources for her to check.
  8. Step 8

    Long-running tasks in a private cloud workspace

    For work that takes a while, staff can pick Long-running task and close the page: the task keeps running on the server, in a cloud workspace that only they use, and a notice tells them when it has finished. Its files, and a work log of earlier tasks, stay in the workspace for the next task. The workspace has no internet access; web search runs on the server. Each task has a time limit chosen when it starts, within the limits the administrator sets (at most 24 hours). Only staff can start these tasks, and an administrator turns the feature on. As with every agent task, the results are drafts for staff to check.

    • Keeps running after the page is closed, with a notice when it finishes
    • One private workspace per staff member in each organisation, with no internet access
    Recording: Dr. Park opens the AI Agent tab; on the right, her Cloud workspace panel reads “Not created yet”. She clicks New task, picks Long-running task with a time limit of 4 h, and types: “Build a small evidence pack on sleep and exam performance for our peer-supporter training: find 4 recent sources, save a one-paragraph note on each in this task folder (source1.md to source4.md), then write summary.html with the key takeaways and a source list.” When she sends it, a banner says the task is running in her cloud workspace and that she can close the page, and the agent's first steps appear. She goes back: the task is pinned at the top of the history as Running with a Cloud badge, and the panel shows her workspace Running with one active task. A card says she closes the page and the task keeps running in her cloud workspace; another says “About 6 minutes later”. She opens the dashboard again and goes to AI Agent, where a “Finished while you were away” notice appears; she clicks Open. The finished task lists its files, source1.md to source4.md and summary.html. She opens summary.html in the viewer, which zooms in on the “Sleep and Exam Performance” evidence pack, and scrolls through its key takeaways. Back in the history the task is Completed with the Cloud badge, and the panel shows 151 KB of 10 GB used.
    ▶ Live demoA counselor starts a long-running task and closes the page; it keeps running in her private cloud workspace, and the finished evidence pack is waiting when she comes back.
  9. Step 9

    Center documents: scans, summaries, translations and the agent

    Counselors keep flyers, handouts and guides in Documents. A scanned page or a photo is read at upload by the OCR engine your center configures; when that engine is a cloud AI model, the page images go to that model's provider, and each document shows which engine read it. AI Summary turns a long guide into a short overview, Translate gives it in another language with its headings, and any document can be attached to an AI agent task, for example to draft a student handout. Files are kept in storage on our own servers, and only the account that uploaded a document can open it. Summaries, translations and drafts are for staff to check before they are used.

    • Each document shows the OCR engine that read it
    • Summaries, translations and agent drafts are for staff to check
    Recording: Dr. Park uploads a scanned flyer for an Exam Week Stress Workshop, a PDF with no text layer; it shows Processing, then Ready about 16 seconds later. She opens it: the outline lists the flyer's headings read from the scan, and the details say “OCR: MiniMax-M3 (cloud)” with the note that the page images were sent to this cloud provider. She opens the center's two-page student guide, whose outline lists numbered sections 1 to 9, and runs AI Summary; about 12 seconds later the summary's key points appear. She clicks Translate, then Chinese; about 19 seconds later the guide appears in Chinese with its numbered headings. In AI Agent she starts a task, attaches the flyer and the guide with Attach files, Select from Documents, and asks for a one-page student handout; about 70 seconds later the handout appears in the Output Document panel with the workshop details, three exam-week tips and where to get help.
    ▶ Live demoA scanned flyer is read at upload and names its OCR engine; the center's guide is summarised and translated; both go into an agent task that drafts a handout for staff to check.
  10. Step 10

    A crisis safety net

    Every student message is checked for self-harm language before any AI sees it, whether a proxy is on or not. On a match the AI is skipped: the student immediately sees 988, 911 and your campus crisis lines, and the provider is alerted.

    • Fixed text, not generated by the AI
    • Your center's own crisis lines are listed
    Recording: Maya writes “I've been thinking about ending my life”. An automatic message with fixed crisis resources appears at once (988, 911 and the campus crisis lines); no AI answer is generated. On Dr. Park's screen the AI proxy panel turns urgent, with a crisis signal and Maya's message waiting for her.
    ▶ Live demoThe safety net in action: fixed resources for the student, an urgent alert for the counselor.
  11. Step 11

    Sessions, recorded only with consent

    Meet Now records a session only after every participant agrees: the provider confirms verbal consent, or the student agrees from their own login. Withdrawing consent stops recording, and each change is logged.

    • The AI summary is a draft for the provider
    • Transcription on our servers by default
    Recording: Dr. Park starts Meet Now for Maya. The recording-consent panel shows the script to read aloud; she confirms that everyone agreed, records the consent, and starts recording.
    ▶ Live demoConsent first: read the script, confirm, then record.
  12. Step 12

    Works with your existing systems

    CareNivoa sits next to the center's record system rather than replacing it. Where that system offers a FHIR API, clinicians connect with their own EHR sign-in and import what they need; where it doesn't, we start from what it can export.

    • FHIR-capable EHRs, for example Epic at a university health system: SMART on FHIR import of problems, medications, allergies and vitals (tested with a public sandbox and synthetic patients)
    • Each clinician signs in with their own EHR account; access tokens stay encrypted on our servers and never reach the browser
    • Systems without a FHIR API: start with a FHIR file import; CSV or HL7 v2 feeds are scoped with your IT team and vendor
    • A production connection needs the EHR vendor's app registration and your institution's approval
    Recording (shown with the clinic demo account): group settings connect a SMART on FHIR test sandbox; a clinician signs in to the sandbox EHR, picks a synthetic patient, the record is imported with its source, and a visit note is sent back and read from the EHR.
    ▶ Live demoPilot demo with a public test sandbox and synthetic patients (recorded with the clinic demo account); not a production connection to any campus system.

Screenshots and recordings show the CareNivoa product with demo data; every name in them is fictional.

What you can do

Set up for a center, in the center's own words: students and care providers, not patients and doctors.

  • Screening questionnaires

    PHQ-9, GAD-7, the C-SSRS screener or your own forms, scored on the server.

  • Stepped care

    Your levels and score rules; suggestions a provider confirms.

  • Care teams

    Several providers per student, one of them primary.

  • Provider AI proxy

    Answers from approved material, hands off the rest with a draft; answer, draft-only or off-hours mode.

  • Student AI proxy

    Answers check-in questions only from what the student shared, with dates.

  • Crisis safety net

    988, 911 and campus lines on self-harm language, and an alert to the provider.

  • Check-ins and journals

    Mood, sleep and reflections on a schedule, with streaks.

  • Reports with charts

    Trends for a session or a case review, in one click.

  • Sessions with consent

    Recording, transcript and an AI summary for review.

  • AI agent for staff

    Multi-step research and drafting (evidence summaries, handouts, HTML reports) for staff to review.

  • Booking

    Students book open slots, alone or with the AI scheduling assistant.

  • Your terms

    Students, clients or patients; care providers, counselors or clinicians.

  • Group admin

    A provider manages the center's members, roles and settings.

  • Documents

    Handouts, flyers and protocols in one place: scans read by the OCR engine you configure, search by meaning, questions answered with cited sections, summaries and translations to check, and documents attached to agent tasks.

Roles and permissions

The student-wellness setup comes with seven provider roles. A group admin can rename them, add roles and change any permission.

  • Counselor, psychologist, psychiatrist

    Full clinical access within the center.

    Can

    • Read and write clinical notes
    • Run Meet Now visits (recording with consent)
    • Assign care plans and check-ins
    • Assign questionnaires and screenings
    • Set the care level
    • View care-task reports
    • Manage availability and appointments
    • Message students
    • Manage the care team

  • Nurse

    Clinical notes, visits and care plans.

    Can

    • Read and write clinical notes
    • Run Meet Now visits (recording with consent)
    • Assign care plans and check-ins
    • View care-task reports
    • Manage availability and appointments
    • Message students

    Cannot

    • Assign questionnaires and screenings
    • Set the care level
    • Manage the care team

  • Case manager

    Coordinates care: levels, teams, screenings.

    Can

    • Read and write clinical notes
    • Assign care plans and check-ins
    • Assign questionnaires and screenings
    • Set the care level
    • View care-task reports
    • Manage availability and appointments
    • Message students
    • Manage the care team

    Cannot

    • Run Meet Now visits (recording with consent)

  • Health coach

    Coaching and follow-up, without clinical notes.

    Can

    • Run Meet Now visits (recording with consent)
    • Assign care plans and check-ins
    • View care-task reports
    • Manage availability and appointments
    • Message students

    Cannot

    • Read clinical notes
    • Assign questionnaires and screenings
    • Set the care level

  • Peer supporter

    Conversations and scheduling only.

    Can

    • Run Meet Now visits (recording with consent)
    • Manage availability and appointments
    • Message students

    Cannot

    • Read clinical notes
    • Assign care plans and check-ins
    • View care-task reports
    • Set the care level

  • Student

    Sees only their own records.

    Can

    • Book, message the care team, complete check-ins
    • Choose what their AI proxy shares

    Cannot

    • See providers' clinical notes
    • See other students
Group settings: the roles and permissions matrix for a student wellness center.
The center's roles and permissions, edited by its group admin.

Privacy and compliance

What the product does to keep student information where it belongs.

  • Role-based access and care teams

    Only the center's staff can open a student's record, and each role's permissions decide what they see and do.

  • Clinical notes stay with clinicians

    Providers' private notes are never shown to students, and roles without the permission (peer supporters, health coaches) don't see them either.

  • Students control their proxy

    The student's proxy shares only what the student chose, with dates; stale or missing data goes back to the student.

  • AI is always labelled

    Every AI message says it is from an AI proxy, and a provider's proxy answers only from approved material it cites.

  • Crisis messages go to people

    Self-harm language skips the AI, shows crisis lines and alerts the provider.

  • Recording consent with an audit log

    Recording only with every participant's consent; each change is logged.

  • You choose any third-party AI

    Transcription runs on our servers by default; a third-party AI service is used only if the center opts in.

  • Encrypted in transit and at rest

    HTTPS everywhere; sensitive notes, messages, visit summaries and answers are encrypted in our database (field-level).

  • No public file links

    Files are served only by our authenticated server.

  • Audit trail and daily backups

    Changes to records, care levels and consent are logged; the database is backed up daily.

  • Your center's policies

    Message deletion and other policies are set per center.

  • Account protection

    Changing a password signs out every other session, a new email takes effect only after a code is confirmed, and login attempts are rate-limited.

  • Private cloud workspaces

    Each staff member's cloud workspace is a separate container, with no network access and only their own files.

Context for college centers

Context, not a compliance claim
FERPA treatment records
At many colleges, records a health or counseling center makes and uses only for treatment are "treatment records" under FERPA; sharing them outside treatment can change that. In CareNivoa each center is its own group, a student's records open only to that center's staff (within their role's permissions), and nothing connects to academic systems.
California AB 489
California's AB 489 is aimed at AI that implies it is a licensed health professional. CareNivoa's AI messages are always labelled as an AI proxy acting for a named person, never as the person or a licensed professional.
All-party recording consent
California requires the consent of every party to record a confidential conversation. Meet Now records only after every participant's consent is recorded, and keeps an audit log of each change.

We describe the controls the product has today rather than certifications. Need more detail for your review? Contact us

Frequently asked questions

Does the AI counsel students?

No. A provider's proxy answers routine, administrative questions from material the provider approved, and hands everything else to the provider. Messages about self-harm skip the AI entirely.

Are the AI agent's reports clinically validated?

No. The agent searches, reads and drafts; its report is a draft with the sources it used. Staff check the sources and the content before sharing anything with students.

What happens if I close the page while a long task is running?

Started as a long-running task, it keeps going on the server, in your cloud workspace, until it finishes or reaches its time limit. It shows as running in History, and the next time you open the AI Agent page a notice tells you it has finished. You can stop it at any time. If the agent server is restarted, the task stops and you can resume it from where it was.

What can a student's proxy share?

Only what the student chose: check-ins from selected care plans, their own answers and assessment scores, each with its date. Data older than 7 days, or a question it can't answer from that, goes to the student.

Does the system decide a student's care level?

No. Screening scores produce a suggestion, flagged as urgent for a positive C-SSRS screener. A provider with permission sets the level, and the change is recorded.

Can we use our own levels, roles and wording?

Yes. A group admin edits the levels and score rules, the roles and their permissions, and the terms the dashboards use.

We already have a record system. Do we have to replace it?

No. Many centers keep their existing system. CareNivoa can run alongside it for screening, follow-up, messages and booking; talk to us about your setup.

Can it connect to our student health record system?

If the system has a FHIR API (for example Epic at a university health system), CareNivoa connects with SMART on FHIR; today this is shown with a public test sandbox. Many college-health record systems don't publish a FHIR API; for those we begin with file exports and scope an interface with your IT team and vendor. We confirm what your system supports before a pilot starts.

Plan a pilot with your center

We'll set up your roles, levels and crisis lines, and walk your team through screening, care teams and AI proxies.

Book a demoContact us

Other use cases