Products · #22 · Documents & Language
TIER S BLUEPRINT

Medical Voice-to-Document (LT)

The doctor speaks, the visit note writes itself - in Lithuanian, structured for e-sveikata, signed by the doctor. Nothing auto-files.

from €1,990

CONSULTATION - AMBIENT CAPTURE · 09:42

Dictation - LT medical speech
CONSENT LOGGED 09:41 ✓EU-ONLY PROCESSING

00:12 "Pacientė ████ skundžiasi galvos svaigimu tris dienas…"

00:42 "…arterinis kraujo spaudimas šimtas penkiasdešimt su devyniasdešimt penki…"

01:17 "…skiriu amlodipiną penkis miligramus kartą per parą, kontrolė po dviejų savaičių."

Visit note - draft · structured for e-sveikata

NUSISKUNDIMAI
Galvos svaigimas, 3 paros ⌕ 00:12

OBJEKTYVIAI
AKS 150/95 mmHg ⌕ 00:42

DIAGNOZĖ (TLK-10)
I10 - pirminė hipertenzija SIŪLOMA - gydytojas tvirtina

PASKYRIMAI
Amlodipinum 5 mg 1×/parą ⌕ 01:17

KONTROLĖ
Po 2 savaičių, AKS dienynas ⌕ 01:17

consent ✓ logged PII redacted ✓ TLK-10 terminology ✓
🟡 GYDYTOJAS PERŽIŪRI, TAISO IR PASIRAŠO - edit distance logged. Never auto-filed.

→ e-sveikata (ESPBI IS) · filed AFTER signature · audit chain ⛓

today ▸ 22 notes drafted · ~47 min saved / clinician · 0 auto-filed

How it works

01

Consent first, then capture

The consent flow is designed with your front desk and logged before a single word is transcribed. Ambient or dictated - audio never leaves the EU.

02

Lithuanian medical speech, understood

LT-tuned recognition: drug names, declensions, specialty jargon, numbers said aloud. Measured on your clinicians' real audio before any number is agreed.

03

Structure, not soup

Mapped to Lithuanian clinical note sections with TLK-10 codes suggested - every field cites the audio timestamp it came from.

04

Doctor signs, system learns

The doctor reviews, edits and signs; edits feed the model. E-sveikata write-back happens only after signature. Falling edit distance is the product working.

The Four Guarantees™ - this build

Measured value

~50% documentation-time cut at reference volume - the agreed number (conservative), gated on a pilot week with your clinicians' real audio.

Defensible

Consent log, PII redaction, EU-only processing, full dictation-to-signature audit chain; TLK-10 suggestions marked as suggestions - the doctor confirms.

Self-correcting

Doctor edits retrain per specialty and per clinician; vocabulary grows with your practice; draft-to-signed edit distance reported monthly.

Yours & everywhere

Your clinic's infra or our EU managed service with per-clinic isolation. Full source. MCP endpoint for your clinic systems.

The number, sized honestly

Reference buyer: Lithuanian private clinic or chain - 15 clinicians, ~20 consultations a day each, 4-6 minutes of documentation per visit, e-sveikata obligations either way.

~50% documentation-time cut - the agreed number (conservative)
45-60 min back per clinician per day at reference volume
0 auto-filed notes - the doctor signs everything, by design
~1-2 mo payback at the PoC tier

Three ways to own it

Tier What you get Price
Scaffolding The full repo - LT medical speech pipeline, note structuring, e-sveikata form library, approve-gate UI, audit chain, MCP server. Reference run on synthetic dictations only. €1,990
PoC ★★RECOMMENDED A pilot on your clinicians' real audio (consent flow included) - transcription accuracy and draft-note edit distance measured per specialty. Code + quality report yours. No guarantee before your audio proves it. from €6,000
Implementation ★★★ Production: clinic-wide rollout, doctor sign-off gates enforced, e-sveikata write-back after signature, per-clinician adaptation, monthly edit-distance + time-saved report - the agreed number (conservative) attaches here. from €18,000

★ = engagement depth. PoC is the recommended path: quality proven on your data before production money. The PoC carries no performance guarantee by design; the agreed number (conservative) attaches at Implementation, informed by the PoC report.

What we don't promise

We don't diagnose, and we don't auto-file. A vendor promising fully automated filing into e-sveikata is selling you a medical-legal incident with a subscription. We measure speech recognition on your clinicians' real audio before any number is agreed - accents, noise and jargon included - and we surface the ESPBI IS integration paperwork as the timeline risk it is, instead of hiding it. The monthly report shows the draft-to-signed edit distance; that number falling is what you're paying for.

Ready to see your own number?

Request the build: within 48h you get a personal reply with the value sized to your volume.

No commitment · reply within 48h · your data stays in the EU