Data warehouse integration

Documents to a Fabric Lakehouse, ready for Power BI

Holofin reads PDFs and scans and writes every field into a Delta table in your Lakehouse, queryable from the SQL analytics endpoint and Power BI.

Try Holofin free Book a demo

Greffe du Tribunal de commerce de Lyon

EXTRAIT KBIS

Code de vérification : 4F7KQ2

DénominationVerger Numérique SAS

Forme juridiqueSociété par actions simplifiée

Capital social50 000,00 euros

Adresse du siège12 quai Saint-Vincent, 69001 Lyon

PrésidentInès Fabre

Extrait délivré le02/06/2026

SELECT field, value, confidence, page, needs_reviewFROM HolofinLakehouse.extracted.document_fieldsWHERE document_id = 'doc_5KF0WD'ORDER BY page, field;

	field	value	confidence	page	needs_review
1	company_name	Verger Numérique SAS	0.99	1	0
2	legal_form	SAS	0.98	1	0
3	share_capital	50000.00	0.97	1	0
4	registered_office	12 quai Saint-Vincent, 69001 Lyon	0.96	1	0
5	legal_representative	Inès Fabre	0.97	1	0
6	extract_date	2026-06-02	0.98	1	1

✓6 fields loaded

!1 flagged for review

A company registry extract read by Holofin and loaded into Microsoft Fabric. Sample document; every name and number on it is invented.

How it works

1
Connect Microsoft Fabric once
Create a role with insert rights on one schema and paste its credentials into Holofin. Holofin creates the table on the first load; nothing else in Microsoft Fabric is visible to it.
2
Send documents the way you already do
Upload in the app, forward by email, or post to the API. Holofin classifies each file, splits bundles into documents and reads the fields of each type.
3
Query the rows
Each field lands as one row with its confidence, the page it came from and a needs_review flag. Join it to your own tables in Microsoft Fabric like any other source.

How the Microsoft Fabric connector writes

Holofin writes Delta tables to OneLake, in a Lakehouse of the workspace you choose. It signs in with a Microsoft Entra ID app registration that you add as a Contributor on that one workspace. The table appears in the SQL analytics endpoint and in Direct Lake semantic models.

The only grant Holofin needs:

# Entra ID app registration: holofin-loader
# Workspace Finance-Ops > Manage access > Add people or groups
#   holofin-loader  ->  Contributor
# Tables land as Delta in OneLake: Files/ and Tables/extracted/

The table Holofin writes

One row per field, so a new document type never changes the table. The same columns exist in every warehouse; types follow Microsoft Fabric's own names.

HolofinLakehouse.extracted.document_fields
document_id	STRING	Holofin's identifier for the document, stable across reloads.
document_type	STRING	The class Holofin gave the document: supplier_invoice, payslip, bank_statement, and so on.
field	STRING	The field name, in snake_case, as defined in your extraction schema.
value	STRING	The value as normalized text: ISO dates, dot decimals, IBANs without spaces.
confidence	FLOAT	Between 0 and 1. How sure Holofin is of the value.
page	INTEGER	The page of the source file the value was read on.
bbox	ARRAY<FLOAT>	Where on that page: x, y, width, height, as fractions of the page.
needs_review	BOOLEAN	True when the confidence is low or a rule failed, such as a balance that does not add up.
reviewed_by	STRING	Who confirmed or corrected the value in Holofin, once someone has.
loaded_at	TIMESTAMP	When the row was written. Corrections write a new row; the latest wins.

Query it in Microsoft Fabric

Find the company extracts older than three months, in T-SQL on the SQL analytics endpoint:

-- Company extracts older than three months, newest supplier first
SELECT TOP 50
  d.document_id,
  n.value AS company_name,
  CAST(d.value AS date) AS extract_date
FROM HolofinLakehouse.extracted.document_fields AS d
JOIN HolofinLakehouse.extracted.document_fields AS n
  ON n.document_id = d.document_id AND n.field = 'company_name'
WHERE d.field = 'extract_date'
  AND CAST(d.value AS date) < DATEADD(month, -3, CAST(GETUTCDATE() AS date))
ORDER BY d.loaded_at DESC;

Security

Documents are processed in Europe, on OVHcloud servers in France certified ISO 27001.
Holofin only ever writes to the schema you grant. It cannot read your other tables.
Retention of the source files in Holofin is set per organization, from zero data retention to indefinite. Rows in your warehouse follow your own policies.
GDPR: Holofin acts as your processor. No document leaves the European Union.

Frequently asked questions

Can I build a Power BI report on the table?

Yes. Add it to a semantic model in Direct Lake mode and the report reads the Delta files in OneLake directly, without an import refresh.

Why is a recent value flagged for review?

Rules can flag a value as well as a low confidence. On a company extract, Holofin flags an extract date more than three months old, since most onboarding procedures ask for a recent one.

How quickly do fields appear in Microsoft Fabric?

Within seconds of the document being read. Most documents are read in under a minute, so a file sent by email is usually queryable in Microsoft Fabric a minute later.

What happens when a reviewer corrects a value?

Holofin writes a new row for that field with the corrected value, needs_review set to false and reviewed_by filled in. Keep the latest row per document and field to get the current state, or keep every row for a full audit trail.

Can I choose which fields are loaded?

Yes. Each document type has an extraction schema in Holofin. Add, rename or remove fields there and the next documents load with the new names. Nothing in the table structure changes.

Which documents does Holofin read?

Any business document sent as a PDF, a scan or a photo: invoices, payslips, bank statements, tax notices, company extracts, IDs, certificates. Holofin also checks each file for tampering and can flag a suspicious document before its fields are trusted.