marzapower

For psychologists and psychotherapists

Emovia

A clearer picture before the next session.

Emovia helps people capture experiences between appointments. Its Interrogative AI asks focused questions; the psychologist receives organized material to explore together in the next session. Clinical interpretation stays with the professional.

Visit emovia.comHow it started

The person writes in the journal

Mon, 22:14

“I barely slept last night, I kept thinking about tomorrow's meeting.”

What about the meeting worried you most?

“Being judged. The idea that every word of mine gets weighed.”

Did that feeling show up tonight, or was it with you during the day too?

The exchange shown on the Emovia homepage, reproduced with the product's own styles. Demo text, not patient data. Source: emovia.com.
Role
I designed the product and its central constraint: the AI that talks to the patient asks questions, it doesn't do therapy.
Status
Evolving. In use in clinical practices, built with psychologists.
Built with
Ruby on Rails, PostgreSQL, React Native; LLM behind the Interrogative AI. Hosted in Europe, encrypted data.
Updated
September 2026

Where it started

Between one session and the next, a patient forgets, or doesn't find a way to hold onto what's happening. At the next session, the professional often starts by reconstructing context instead of building on it. According to data Emovia collected, 37% of people already use general-purpose chatbots for emotional support, with no clinical supervision at all.

Emovia was built to close that gap with a tool designed for therapy, not a generic chatbot: it gives the patient a familiar journal and chat to write down thoughts and feelings between appointments, and gives the professional that material back, already organized.

It was co-founded with a psychologist, Elisa Manca, and built together with a group of psychologists from the earliest decisions: what the AI can ask, what it can surface, where its boundary ends.

What it does for clinicians and patients

Six things, split between the patient's side and the professional's.

  1. 1

    A journal and a chat for the time between sessions

    Between appointments, the patient writes in the journal or opens the chat to note a thought, a feeling, an event — the same kind of tools they already use every day, applied to the therapeutic process.

  2. 2

    The Interrogative AI only asks questions

    The assistant that talks to the patient is scoped to ask targeted questions, never to interpret or advise: it helps them focus on what they're writing, without doing therapy in your place.

  3. 3

    A dashboard that summarizes the period

    Before the session, the professional finds a summary of the period, the recurring themes that emerged from the journal, and the key events — instead of having to reconstruct all of it out loud.

  4. 4

    Clinical supervision in four steps

    Every conversation moves from an automatic check to an alert for the safety team, up to blocking the chat and returning it to the professional if something falls outside the expected boundaries.

  5. 5

    The professional decides who gets in and what the AI does

    The psychologist activates access, sets the instructions, and governs what the assistant can ask their own patient: control stays clinical, it isn't delegated to the software.

  6. 6

    Clinical data treated as such

    GDPR compliance, hosting in Europe, encrypted data, and no model training on patient content: confidentiality is a design condition, not a promise.

How it's built

For anyone who wants to look underneath.

Before writing a line of code, I decided the boundary of the AI that would talk to the patient: it can ask questions, it can't interpret or advise. That choice shows up in every prompt and every guardrail built after it, and it's what lets a practice adopt Emovia with confidence, knowing exactly what the AI can and can't do.

Patient data and professional data live on separate surfaces from the code up: different access, different purposes. All of it on hosting in Europe, with data encryption and role-based access control, so clinical information stays where it needs to stay.

Where it stands

Emovia is in production, in use in clinical practices, and clinical validation continues with the group of psychologists it was built with.

Evolving, September 2026

Bringing AI into a domain where trust matters?

I designed Emovia to stay inside a precise boundary, not to push it. If you're facing a similar problem, let's talk.

Get in touch

More work

  • Fabulous FactoryFoundations already in place for your AI agent. It builds the product.
  • AriannaStudy from a manual built for you, not from scattered slides.
  • Power PuzzlesPuzzles you solve by reasoning, never by guessing.