Mentavita is an AI psychologist with memory, voice, and a journal, plus a platform for working with verified psychologists — in one app on Android, iOS, and the web.
3
platforms: Android, iOS, and web — Android is the reference, iOS and web match it
2
paying sides: clients for depth of AI support, psychologists for practice tools
0
critical or high findings in the full security audit of Oct 19, 2026 across six areas
1.1 ms
per database query after the move to Europe on Sep 30 (was 37.7 ms)
> 1 bn
people worldwide live with mental health conditions
2%
median share of government health budgets spent on mental health
13
mental health workers per 100,000 people — global median
Source: WHO, “Over a billion people living with mental health conditions”, Sep 2, 2025 (World mental health today, Mental Health Atlas 2024).
Survey of Nov 12, 2022, 1,600 respondents aged 18+. “Other answers” is the remainder to 100%.
Source: VTsIOM, as reported by Vedomosti on Nov 22, 2022.
Market sizing for Russia and the CIS and its assumptions are in the investment memorandum.
One loop from daily support to a live specialist. Roles are separated: a psychologist sees client data only within their work together and under the client’s consents.
24/7 chat, voice conversations, and a journal — the person controls AI access to it.
A graph of events, people, and reactions: the AI remembers what matters, the person sees and edits it.
A journal draft, a practice, or an experiment — one at a time; the radar suggests a pause in advance.
Matching in 4 questions, a catalog of verified specialists, booking, and a video session.
An AI assistant for the psychologist, a pre-session summary, and the client’s AI within the psychologist’s approach.
A crisis protocol at every step: deterministic message checks and an SOS panel with 112 and helplines.
24/7 chat on a modern language model. Memory v2 is a graph of events, people, and reactions: the AI follows up on what matters, and the person can ask it to forget.
A real-time voice conversation, with a summary and transcript afterwards and minutes per plan. Currently a pilot with a test group.
Recurring thoughts and reactions — first as a “hypothesis” to confirm; weekly reports and “Your week”.
Entries, photos, voice notes with transcription, and mood. AI access to the journal can be turned off in settings.
“Solomon’s mirror”: the person advises themselves as a friend, the AI runs the rounds and sums up. Not offered after recent signs of crisis.
A journal draft, a practice from the catalog, an experiment, a gentle bridge to a psychologist — only when relevant; the dialogue stays first.
With consent and from numeric signals only (sleep, rhythm, mood), suggests a pause in advance. No pushes, at most weekly; rolled out in stages.
Mood in half a minute; the AI follows up on important events — at most twice a week and never in quiet hours.
Rehearse a difficult conversation with the AI before the real one.
10 community achievements for clients and 10 for psychologists — shown at the person’s choice.
A profile after document verification, a “Verified specialist” badge; clients find psychologists by matching in 4 questions. Catalog order is the same for every plan.
Day and month views, drag-to-reschedule, custom services and group sessions, sync with external calendars.
Video on our own infrastructure; a guest link with a meeting ID and code — for example, for the second partner in couples work.
Only with client consent. After the session — a SOAP / DAP note draft that the psychologist edits and approves.
Answers only about this psychologist’s clients and only with facts linked to sources; says so when there is no data.
With client consent, their AI chat follows the psychologist’s approach, goals, and focus; a summary is prepared 3 hours before the session.
From the chat or after a session; AI helps draft it, and the client gets reminders.
An AI assistant finds answers in psychotherapy books and articles and shows the source. If there is no answer, it says so.
Bilateral stimulation during an online session — for specialists with verified training.
Major releases of recent weeks — on Android, iOS, and the web unless noted otherwise. Dates come from the project’s engineering log.
A new language model for chat and reports, a scribe for psychologists (recording, transcripts, notes), client–psychologist chat with files.
PRO / PRO+ and Start / Practice / Pro plans, founders program; psychologist journal; check-in, “AI remembers”, “Your week”; QR sign-in; iOS caught up with Android.
Servers, database, and video infrastructure in a European data center: 1.1 ms per database query instead of 37.7. New AI memory.
Answers for psychologists about their clients with source links; the client’s AI within the psychologist’s approach.
A link, meeting ID, and access code for a participant without an account.
Event-centric memory: people, time, place, reactions; forgetting propagates along links.
Psychologist catalog, matching in 4 questions, “Verified specialist”, custom services and group sessions, card-linked PRO+ trial.
Round video messages in chats, offline media encryption on Android.
Real-time voice conversation with the AI, “Outside view”, achievements, live streams, client referrals, radar in shadow mode.
Minutes per plan, minute packs, “PRO+ Voice”; a full security audit across six areas.
Recording, guests, co-hosts, moderation, polls, Q&A, booking from a stream, PiP; timely suggestions in chat and voice; radar stage 2.
Start — 1 stream a month, Practice and Pro — unlimited.
An entry is a finished block of work on one platform: a feature, an audit, or a round of fixes. The log starts on Sep 25, 2026; the last week is partial (as of the morning of Oct 21).
Source: Mentavita project log, 302 entries.
Counts the modules described on this page. All are available on Android, iOS, and the web; Android is the reference client.
Source: product description on this page, as of Oct 21, 2026.
Clients pay for depth of AI support and voice minutes, psychologists for practice tools. Card payments on the web and in the apps, renewals charged by our own backend; monthly or annual.
5 AI messages a day, a 5-minute trial voice conversation
20 voice minutes a month
60 voice minutes; 7-day card-linked trial
150 voice minutes, staged rollout
Prices as of Oct 21, 2026; current ones are on the Pricing page. Minute packs: 290 ₽ for 30 min and 790 ₽ for 100 min, valid 90 days — staged rollout. Pricing
Source: Mentavita plan catalog.
up to 3 active clients, 1 stream a month
first 50 psychologists: 3 months free, then for life
1,908 ₽ a month when paid yearly
3,741 ₽ a month when paid yearly
Source: Mentavita plan catalog, founders program.
Measured on real calls: a minute of voice conversation costs 2.3–2.8 ₽ (with context compression and a compact model instruction). Below is a calculation: what share of a plan’s price voice takes if every minute is used, and the margin minute packs leave over voice cost.
20 min × 2.3–2.8 ₽
60 min × 2.3–2.8 ₽
150 min × 2.3–2.8 ₽
Source: cost measurement on real calls, October 2026; minutes and prices from the plan catalog.
9.67 ₽ per minute
7.90 ₽ per minute
A calculation, not reporting. Solid part — at 2.3 ₽/min, hatching — up to 2.8 ₽/min. Text AI, acquiring, taxes, and infrastructure are excluded; unused minutes lower the actual share.
Source: cost measurement on real calls, October 2026; minutes and prices from the plan catalog.
We don’t publish user and financial metrics on a public page. They are shared on request together with the memorandum.
MAU / WAU, share of voice users — on request.
Breakdown by plan and side — in the investment memorandum.
Trial → paid, cohort retention — on request.
Channels and acquisition cost (CAC) — on request.
Deterministic crisis checks in chat, voice, Mirror, live streams, and community → an SOS panel with 112 and helplines. A crisis bypasses limits: a voice call is extended beyond the minutes.
The AI does not diagnose or replace a specialist — and says so. In a crisis the AI switches to crisis mode, Mirror is not offered, and reminders are paused.
Separate revocable consents: session recording, summary for the psychologist, tandem, voice tone analysis, radar, AI access to the journal. AI memory controls and data export.
Soft deletion and 3-year retention after account deletion, a personal data registry in code, an access log for sensitive data. Rule: texts, mood, and crises stay out of analytics — audits check it.
Regular audits (Sep 25, Oct 1, Oct 12, Oct 19) against OWASP ASVS, MASVS, and LLM Top 10. The full audit of Oct 19 found no critical or high issues; the rest is fixed on schedule.
Psychologists edit and approve AI notes; AI moderation only flags — moderators decide; radar templates are reviewed by a psychologist.
The team brings together product specialists, engineers, and psychological counselors. More on the team page.
About the team8+ years in digital products: launching and growing web/mobile SaaS, product strategy, growth, and user research.
LinkedInFull-stack engineers: AI integrations, infrastructure, product analytics, performance, and security.
Company LinkedInRoles, commitment, and cap table — in the investment memorandum.
Open voice conversations, “PRO+ Voice”, and minute packs — all built and switched on by flags after cost measurements.
Team → 5% of users → everyone, after a psychologist’s review and forecast accuracy thresholds.
Larger video servers, stream recording separately; a containerized backend behind a load balancer, a managed cloud database.
Group work with one psychologist and strict isolation of each person’s data; several psychologists per client.
Late arrivals, late cancellations, and no-shows — automatic and with the psychologist’s rules.
Psychologists’ courses and practice analytics for Pro — marked “coming soon” in the plans.
We will send the deck and the investment memorandum with metrics and a financial model, and give a live product walkthrough. Write by email or via the contact form.
invest@mentavita.app