A reference build: we made this ourselves to show how we work. It is not a client engagement, there is no customer to name, and there are no measured outcomes to report - so this page does not pretend to have any.
Three legal entities, one price list, no reconciliation night
A reference build: billing across several entities and insurers, transmitting into myDATA through a certified provider, with the service and VAT mapping treated as the actual project.
- Client
- Reference build
- Sector
- Multi-site clinics
- Year
- 2026
- Built with
- Next.js, Postgres, myDATA, Python
The numbers
- 3
- Legal entities with one price list in practice and three in law
- 3
- Things mapped explicitly per site and contract: service code, VAT treatment, issuing entity
- 0
- Documents transmitted before the mapping was reviewed with the group's accountant
The problem
A clinic group operating through three legal entities has one price list in practice and three in law: the same procedure carries a different VAT treatment, a different insurer contract and a different issuing entity depending on where it was performed. Transmission to myDATA is the easy half. The hard half is that a mapping error surfaces as a reconciliation problem months later, by which point hundreds of documents share it.
What we did
- 1
The mapping is the deliverable, not the plumbing. Service code, VAT treatment and issuing entity are modelled explicitly per site and contract, reviewed with the group's accountant before a single document is transmitted, and versioned so a change is dated rather than retroactive.
- 2
Validation happens before transmission, against the schema and against the group's own rules, so a rejection is caught at the point of issue by the person who can fix it - not by ΑΑΔΕ, and not by the accountant in the following quarter.
- 3
Reconciliation runs continuously rather than monthly, and reports mismatches as they appear. A discrepancy found the same day is a correction; the same discrepancy found in March is an audit.
- 4
Access to anything patient-linked is logged on read as well as on write, because health data is special-category and the question a regulator asks is who looked at a record, not who could have.
Where it got to
The change is that the mapping is now a reviewed, dated artefact rather than tribal knowledge held by one person in accounts. No figures are quoted here because none were measured in a way we could show you.
A worked example
Transmission is the easy half. The mapping is the deliverable, and an error in it compounds quietly:
- Same procedure, VAT in clinic vs at home
- differs
- Mapping error found the same day
- a correction
- The same error found in March
- an audit
- Reconciliation
- continuous, not monthly
- Opening a patient record
- logged
What changed is that the mapping is a reviewed, dated document instead of knowledge held by one person in accounts. No figures appear because none were measured in a way we could show you.