~/projects/emovia
Emovia
Emovia is a B2B SaaS for psychologists that gives patients a guided journal between sessions and gives professionals organized clinical material before the next one.
The problem
Between one therapy session and the next, a lot happens — and most of it gets lost. Patients forget details, and professionals often start each session reconstructing context instead of building on it.
Emovia is a B2B SaaS built for the Italian psychology market. It has two sides: a guided, AI-assisted journal for the patient to use between sessions, and a set of clinical workflow tools for the professional.
The goal isn't to replace the therapeutic relationship, and it isn't a consumer chatbot: general-purpose AI used unsupervised in a clinical setting can harm as easily as it helps. Emovia is built as a professional tool where the guardrails on the AI come first — structured for clinical use, not left to a chatbot's defaults.
My role
I designed the product and its central technical constraint: the AI that talks to patients — the "Interrogative AI" — asks targeted questions. It does not interpret, it does not advise, it does not do therapy.
That boundary shaped every decision downstream: what the model is allowed to generate, what the patient sees, and what surfaces to the professional afterward.
The product was built with psychologists, for psychologists — clinical input shaped the workflow tools and the boundaries placed on the AI from the start.
My aim with Emovia is to put my engineering into making AI as safe as it can be in a domain where sensitivity is everything — evidence that strong guardrails and professional-grade AI can coexist.
Architecture & decisions
The engineering problem here isn't making the AI capable — it's making it stay inside its lane, in a domain where a wrong inference has real consequences.
Layered guardrails, not a single switch
The Interrogative AI is scoped to ask questions, not to answer them — a deliberate ceiling, not a current limitation. The guardrails are layered: careful, detailed prompting first; self-supervision built into the chatbot and its AI tools; and a dedicated product-safety team adding both automated and human review on top.
Two sides, one boundary
The patient-facing journal and the professional-facing workflow tools are separate surfaces with different data access and different jobs: one helps someone notice what's happening between sessions, the other gives the clinician organized material to work from. The separation between patient and professional data is enforced in code.
Clinical data as a design constraint
Handling sensitive clinical information shaped the product from the ground up: full GDPR compliance, hosting in Europe, encrypted storage and encryption in transit, and granular, role-based access control (RBAC).
Built with clinicians, not just for them
Clinical validation runs in collaboration with the team of psychologists the project was built with — more about them on Emovia's own site.
Screenshots
Patient journal view
[ASSET NEEDED: Patient-side journal screen with an Interrogative AI prompt]
Professional workflow dashboard
[ASSET NEEDED: Professional-side dashboard showing organized session material]
Session preparation view
[ASSET NEEDED: Pre-session summary view the professional sees before the next appointment]
Status
Emovia is in production and live at www.emovia.com. The roadmap centers on new AI-driven features.
Related writing
Technical write-ups on Emovia are on the way.