Core
Running-record chart
A fixed patient header, the initial assessment and progress notes stacked by date in a single continuous flow. No more opening and closing screens to understand the case — the whole history is in one scroll.
LNData Product · FisioFlux In development · pre-launch
The system that talks to the health professional.
FisioFlux speeds up patient registration and follow-up by talking to whoever runs the session: the professional jots the note down in free text — the way they think — and AI turns it into a structured medical record, always as a draft to review before saving. It is LNData’s clinic-management software for physical-therapy, pilates, osteopathy and acupuncture studios, with the "running-record" chart at its core.
Role · Proprietary product · architecture, full-stack engineering, applied AI and LGPD compliance
Status · Multi-tenant SaaS · in development · pre-launch
In short, for decision-makers
The problem
In clinic-management systems the patient’s history is scattered across screens and tabs — the professional opens and closes windows just to understand the case, and clinical record-keeping becomes bureaucratic work that steals time from care. Add scheduling, attendance and session credits handled separately, and the studio’s operation ends up fragmented.
What LNData is building
FisioFlux unifies scheduling, clinical progress and session credits into a single continuous chronological view — the running record. The schedule generates the dates automatically, credits follow the attendance rule (cancel and you can make up, no-show and you lose it), and an AI talks to the professional to turn free text into a structured record. All multi-clinic and under LGPD compliance for sensitive health data.
The promise
Less time typing and hopping between screens, more time with the patient: record-keeping gets fast, history fits in a single scroll, and studio management — schedule, attendance, credits and finance — moves together. AI speeds things up, but the professional stays in control: nothing enters the record without their review.
Running record
history in a single flow
SaaS
multi-clinic (multi-tenant)
LGPD
health data by design
Pre-launch
in development · v0.11
Text → record
you write, AI structures it
Always reviewable
AI never saves on its own
Auto scheduling
enroll, and the dates already exist
Multi-tenant
schema isolation, per-clinic backup
Architecture
The professional writes the note the way they think; an AI assistant organizes it into the clinical format; and the record receives the result — always as an editable draft. Every external integration (AI, payment, messaging, storage) sits behind its own interface, swappable and testable.
The professional writes
- Free-text note
- The way they think · no rigid form
- Only the text goes to the AI · no patient identifiers
AI assistant
Text → structured SOAP
Tenant-aware MCP server
Providers isolated by interface
Claude (Anthropic) · consent gate
The record receives
- A SOAP draft to review
- Stacked by date on the chart
- Versioned · never deleted, audited
What the platform does
A complete clinic management with the running record at its heart — scheduling, progress, attendance, credits and finance in one place, with AI and WhatsApp as part of the product.
AI
The AI that talks to you
The professional writes the note in free text and the AI proposes a structured record (SOAP). It is always an editable draft — the AI never writes directly, by ethical and legal requirement.
Scheduling
The schedule generates the dates
When a patient is enrolled in a package (e.g. 2×/week → 8 dates), the system creates the chart rows on its own. The professional never types a date. Includes rooms, equipment and group classes with a waiting list.
Attendance
Credit tied to attendance
Session status in one click — attended, cancelled or no-show — with the central business rule: cancel and you can make up, no-show and you lose the credit. No parallel counter to drift.
Relationship
Contact today, via WhatsApp
A daily contact list — reactivating those who dropped off, birthdays, session reminders — ready to fire over WhatsApp, with specific, revocable consent for the contact.
Finance
Billing and recurrence
Payment integration with Pix, recurring card and idempotent webhooks, plus an executive dashboard with the clinic’s KPIs — the financial operation inside the same system.
The intelligence that makes the difference
In a system holding sensitive health data, ungoverned AI is a risk. FisioFlux was designed so AI speeds things up without ever taking control — or the data — out of the professional’s hands.
AI always as a reviewable suggestion
The assistant proposes; the professional reviews and saves. The AI never writes directly into the record — an ethical and legal requirement that became an architecture rule, not a config option.
Data minimization by design
Only the clinical text goes to the AI, never the patient’s identifiers. Consent for AI and for WhatsApp contact is specific, separate from general consent, and revocable — checked before any external call.
Tenant-aware MCP server
A clinical-domain tool server (patient briefing, reactivation candidates) that resolves the correct clinic by credential before any query — each action isolated per tenant and audited.
Beyond the core
Multi-tenancy, LGPD and isolated providers
The running record is the visible part; underneath is the infrastructure that sustains many clinics with real isolation and treats health data as a first-class requirement — the foundation already built and tested.
Physical multi-tenancy by schema
Each clinic is an isolated PostgreSQL schema, routed by subdomain. It allows granular per-clinic backup and deletion (essential for portability and LGPD erasure) without touching the others.
LGPD for health data
Clinical fields encrypted at rest, immutable read-and-write auditing (append-only), a versioned record that is never deleted, and role-based access control — reception does not see clinical data.
Isolated, swappable providers
Payment, AI, messaging and storage talk to an abstract interface, never to the external service directly. Each has a fake provider for offline tests and is swappable without touching the domain.
Tenant-aware consistency
Async queues and the MCP server do not inherit the tenant automatically — a code pattern (`@tenant_task`) forces the explicit schema, closing a classic cross-clinic leak trap.
Djangodjango-tenantsPostgreSQLpgcrypto
The vision
Clinical record-keeping stops being bureaucracy: the professional writes the way they think and the AI structures it, giving time back to care.
Scheduling, progress, attendance and credits stop living in separate screens — the running record brings it all into a single chronological view.
Health data treated as sensitive from the ground up: encryption at rest, immutable auditing and revocable consent for AI and WhatsApp.
Foundation already built and tested (multi-tenant, LGPD, running record, finance, 160+ tests); product in development, heading to launch.
Platform stack
DjangoDRFReactCeleryPostgreSQLMinIOAnthropicMCPDocker
Want to follow FisioFlux closely?
FisioFlux is in active development, heading to launch. If you run a physical-therapy, pilates or related clinic and want a system that talks to the professional and speeds up the record, talk to LNData to follow the launch or join the early tests.