Time
The session is transcribed and the documentation comes out of it. Test scoring, case note and treatment plan arrive as drafts instead of being typed up in the evening.
Practice and clinic software for the Arabic-speaking region
More treatment.
Less administration.
Hauna brings practice management, assessment, therapy tools and AI-assisted documentation into one system — in Arabic, not in translation. The work that piles up after a session is produced during it.
- 140+
- psychometric instruments, scored automatically
- 240+
- therapy tools and six group programmes
- 8,568
- Arabic clinical text blocks — not translated trimmings
- On site
- the language model runs on your server, not in a cloud
What you gain
Four points where a practice's day changes. Each one is a feature in the system — not a projection.
More treatment
Whatever you no longer score or retype yourself remains as treatment time. Online booking and the waiting list fill open slots without anyone making a phone call.
A wider offering
Six structured group programmes and over 240 tools are already there — group and guided self-help formats can be offered without developing them first.
Clean billing
Service capture, billing and invoices in the same system as the record. No transfer between programmes, no second set of data.
We deliberately quote no percentage for time saved. We have not measured it, and an estimate would be a claim. What stands above describes what the software does — verifiable in a walkthrough against your own workflows.
What the system does
Four building blocks, one body of data. What assessment produces lands in the record; what is said in session becomes the draft documentation — no retyping, no second programme.
Administration and record
Appointments, calendar, online booking, waiting list, day list, billing and invoices — alongside intake, course of treatment, findings and letters in one place.
- Online booking and waiting list without the telephone
- Video consultation self-hosted, no third-party provider
- Utilisation and outcome reporting for management
Assessment
Over 140 instruments, from PHQ-9 and GAD-7 to BDI-II, MADRS, PCL-5 and AUDIT. The patient completes them on their own device; scoring appears in the record immediately.
- Scoring, severity and written findings generated automatically
- Response scales and categories in all six languages
- Crisis-relevant answers are detected by rule, not by model
Therapy tools
Over 240 exercises and worksheets, plus six structured group programmes for ADHD, anxiety, depression, couples, social skills and stress.
- Assign to a patient, see the response in the record
- Treatment plans as templates the clinician can adapt
- Content is multilingual, not machine-translated after the fact
AI-native features
The video session is transcribed live; from it comes the draft of the case note and the treatment plan. Approval always stays with the clinician.
- Session documentation and treatment application pre-filled
- Every claim backed by a verbatim quote with a timestamp
- The AI makes no diagnosis and sets no session quota
Arabic is the first language here
It is not the interface that was translated; the clinical content was maintained in Arabic. That is the difference you only notice in daily use.
8,568
Clinical text blocks
Assessments, tools, programmes and treatment plans in Arabic — counted in the source, not estimated.
46
Arabic crisis phrases
Crisis detection works by rule in Arabic and distinguishes levels: an explicit suicidal statement reaches the highest level, a neutral statement none.
45 s
Arabic session analysis
Measured on our server: from an Arabic session transcript, the documentation draft takes 45 seconds — on ordinary CPU cores, without a graphics card.
RTL
Right to left
Writing direction is part of the layout, not a switch. Arabic emergency numbers and regional contacts are configured per site.
What we have not measured is the word error rate of Arabic speech recognition. For German it is 2.6 per cent (our own measurement); for Arabic the evidence is outstanding, which is why no figure appears here.
European standard, verifiable
Health data is a special category under Article 9 GDPR. The architecture treats it that way — not as a setting somebody can forget.
The data stays where it is
The language model runs on your machine. A guard in the code blocks cloud providers wherever patient data could appear in a prompt — it cannot be switched off by configuration.
Tenant separation in the database
Separation between organisations is enforced by the database (row level security), not merely by the application. A coding mistake therefore cannot expose another organisation's records.
Every AI answer leaves a trace
Model, prompt version and timestamp are stored with the result. Without them, nobody could later say how a draft came about.
Two-factor access
Clinical and administrative roles require two-factor sign-in. Particularly sensitive fields are encrypted at field level.
The security architecture is documented in 15 layers, from role checks and prompt-injection detection through to the nightly audit-log review. We provide the documentation on request.
Built multilingual, not translated afterwards
The platform was built in Germany and designed for care across several language regions — that was a requirement, not an addition.
Six languages
German, English, Arabic, Turkish, Russian and French — interface, assessments and therapy content.
Arabic right-to-left
Writing direction is part of the layout, not a switch bolted on afterwards. Arabic content is clinically maintained.
Emergency numbers per region
Crisis contacts are configured per location. Someone in Cairo shown a German number would reach nobody.
Connectable
FHIR export and a connection to the German telematics infrastructure are in place. Billing catalogues come from official sources.
Who it is for
Single and group practices
One piece of software instead of three: record, assessment and video consultation without switching programmes.
Care centres and practice networks
Several sites as separate tenants — shared administration, separated data.
Clinics and outpatient departments
Structured group programmes, a roles and approval model, and an audit trail for internal review.
What the system is not, today
This list is here because you would ask anyway — and because a vendor who hides it is being imprecise about everything else too.
Not an approved medical device
There is no CE marking and no listing as a reimbursable digital health application. Classification under the EU Medical Device Regulation is open and being clarified.
No clinical efficacy study
There is no outcome data from routine care. Maturity is TRL 5–6: fully built, not yet clinically validated.
The AI decides nothing
It produces drafts. Diagnosis, treatment decisions and approval rest with the clinician — enforced in the architecture, not merely agreed.
A walkthrough in your own setting
We show the system against your workflows, not a demo practice. Tell us what you use today and where it gets in the way.
Request a walkthroughnoha.eissa@haunacare.com