Skip to main content
provence.ai
AI platform · regulated sectors

From raw documents to decisions.

Your cases flowing, verified, delivered.

A sovereign AI platform that brings reliability to the processing of your most sensitive documents.

Request a demoExplore the platform

Trusted by

  • CD Landes
  • Vaucluse
  • CNES
  • Digitech
  • Worldline
ISO 27001:2022
HDS compatible
SecNumCloud compatible
Frugal AI
GDPR compliant

The challenges you already know.

Invisible time, massive cost

Your teams spend a huge amount of time reading, sorting and keying information from every single document. It's invisible, but it weighs on everything.

Sensitive data everywhere

Health records, ID documents, proofs of address, bank details. Every document demands hardened protection, and every operation a complete audit trail.

Fraud is industrializing

AI-generated fakes, Photoshop edits at scale, hijacked official codes. Your investigators are running a month behind the fraudsters.

Your users won't adapt. We will.

Blurry photos, cryptic handwriting, IT systems that don't talk to each other. We read, classify, check and deliver - so your teams decide rather than type.

Documents from real life.

Your users send you photos that are blurry, crooked, badly lit.

provence.ai captures the document, extracts and checks the data.

No user should ever bend to your process - the technology adapts to their daily life.

Human writing, decoded.

Handwriting is everywhere, because we are human.

And it often holds critical information that must be read, checked and recorded.

Prescriptions are cryptic. Not us.

Tablette affichant un document en cours de traitement
{
"dossier_id": "DSR-2024-08471",
"date_reception": "2024-11-07T09:41:12Z",
"pages_recues": 14,
"decoupe": {
"pieces_detectees": 6,
"methode": "vision_segmentation_v3",
"resultats": [
{ "index": 1, "type": "piece_identite", "pages": [1, 2], "confiance": 0.997 },
{ "index": 2, "type": "justificatif_domicile", "pages": [3], "confiance": 0.994 },
{ "index": 3, "type": "rib", "pages": [4], "confiance": 0.991 },
{ "index": 4, "type": "bulletin_salaire", "pages": [5, 6, 7], "confiance": 0.989 },
{ "index": 5, "type": "avis_imposition", "pages": [8, 9, 10], "confiance": 0.996 },
{ "index": 6, "type": "attestation_caf", "pages": [11, 12, 13, 14], "confiance": 0.988 }
]
},
"classification": {
"modele": "doc_classifier_v8.2",
"dossier_complet": true,
"pieces_manquantes": [],
"pieces_classees": [
{ "piece": "CNI", "sous_type": "nouveau_format", "recto_verso": true, "score": 0.998 },
{ "piece": "EDF", "fournisseur": "Engie", "date_facture": "2024-10-15", "score": 0.996 },
{ "piece": "RIB", "format": "SEPA", "score": 0.993 },
{ "piece": "BULLETIN_SALAIRE", "employeur": "Sodexo SA", "mois": ["08","09","10"], "score": 0.991 },
{ "piece": "AVIS_IMPOSITION", "annee_revenus": 2023, "foyer_fiscal": "1A", "score": 0.997 },
{ "piece": "ATTESTATION_CAF", "organisme": "CAF13", "score": 0.989 }
]
},
"extraction": {
"identite": {
"nom": "MARTIN",
"prenom": "Isabelle",
"date_naissance": "1987-03-14",
"nir": "2 87 03 75 112 045 62",
"adresse": "12 rue des Lilas, 13100 Aix-en-Provence"
},
"financier": {
"iban": "FR76 3000 4028 3700 0100 0425 082",
"bic": "BNPAFRPPXXX",
"revenus_net_mensuel": 2847.63,
"revenu_fiscal_reference": 34171,
"montant_apl": 312.00
},
"employeur": {
"raison_sociale": "Sodexo SA",
"siret": "301 940 956 00876",
"date_embauche": "2019-06-03"
}
},
"controles_antifraude": {
"moteur": "fraud_engine_v5.1",
"score_global": 0.03,
"verdict": "DOCUMENT_AUTHENTIQUE",
"analyses": [
{
"controle": "coherence_identite",
"statut": "OK",
"detail": "NIR, nom, prenom, date concordants"
},
{
"controle": "authenticite_cni",
"statut": "OK",
"score_falsification": 0.012,
"detail": "MRZ valide, hologramme detecte"
},
{
"controle": "2ddoc_verification",
"statut": "OK",
"detail": "Signature ANTS verifiee"
},
{
"controle": "coherence_adresse",
"statut": "OK",
"detail": "Adresse identique sur 4 pieces"
},
{
"controle": "falsification_pdf",
"statut": "OK",
"score_falsification": 0.008,
"detail": "Aucune couche suspecte"
},
{
"controle": "coherence_revenus",
"statut": "OK",
"detail": "Ecart cumul vs avis: 0.2%"
},
{
"controle": "iban_beneficiaire",
"statut": "OK",
"detail": "IBAN correspond au titulaire"
},
{
"controle": "detection_clones",
"statut": "OK",
"comparaisons": 847291,
"detail": "Aucun clone detecte"
}
],
"temps_traitement_ms": 342
},
"ingestion": {
"systeme_cible": "GED_Alfresco",
"statut": "INGERE",
"dossier_cree": "ALF-2024-08471",
"pieces_indexees": 6,
"metadonnees_injectees": true
}
}
{
"dossier_id": "DSR-2024-08471",
"date_reception": "2024-11-07T09:41:12Z",
"pages_recues": 14,
"decoupe": {
"pieces_detectees": 6,
"methode": "vision_segmentation_v3",
"resultats": [
{ "index": 1, "type": "piece_identite", "pages": [1, 2], "confiance": 0.997 },
{ "index": 2, "type": "justificatif_domicile", "pages": [3], "confiance": 0.994 },
{ "index": 3, "type": "rib", "pages": [4], "confiance": 0.991 },
{ "index": 4, "type": "bulletin_salaire", "pages": [5, 6, 7], "confiance": 0.989 },
{ "index": 5, "type": "avis_imposition", "pages": [8, 9, 10], "confiance": 0.996 },
{ "index": 6, "type": "attestation_caf", "pages": [11, 12, 13, 14], "confiance": 0.988 }
]
},
"classification": {
"modele": "doc_classifier_v8.2",
"dossier_complet": true,
"pieces_manquantes": [],
"pieces_classees": [
{ "piece": "CNI", "sous_type": "nouveau_format", "recto_verso": true, "score": 0.998 },
{ "piece": "EDF", "fournisseur": "Engie", "date_facture": "2024-10-15", "score": 0.996 },
{ "piece": "RIB", "format": "SEPA", "score": 0.993 },
{ "piece": "BULLETIN_SALAIRE", "employeur": "Sodexo SA", "mois": ["08","09","10"], "score": 0.991 },
{ "piece": "AVIS_IMPOSITION", "annee_revenus": 2023, "foyer_fiscal": "1A", "score": 0.997 },
{ "piece": "ATTESTATION_CAF", "organisme": "CAF13", "score": 0.989 }
]
},
"extraction": {
"identite": {
"nom": "MARTIN",
"prenom": "Isabelle",
"date_naissance": "1987-03-14",
"nir": "2 87 03 75 112 045 62",
"adresse": "12 rue des Lilas, 13100 Aix-en-Provence"
},
"financier": {
"iban": "FR76 3000 4028 3700 0100 0425 082",
"bic": "BNPAFRPPXXX",
"revenus_net_mensuel": 2847.63,
"revenu_fiscal_reference": 34171,
"montant_apl": 312.00
},
"employeur": {
"raison_sociale": "Sodexo SA",
"siret": "301 940 956 00876",
"date_embauche": "2019-06-03"
}
},
"controles_antifraude": {
"moteur": "fraud_engine_v5.1",
"score_global": 0.03,
"verdict": "DOCUMENT_AUTHENTIQUE",
"analyses": [
{
"controle": "coherence_identite",
"statut": "OK",
"detail": "NIR, nom, prenom, date concordants"
},
{
"controle": "authenticite_cni",
"statut": "OK",
"score_falsification": 0.012,
"detail": "MRZ valide, hologramme detecte"
},
{
"controle": "2ddoc_verification",
"statut": "OK",
"detail": "Signature ANTS verifiee"
},
{
"controle": "coherence_adresse",
"statut": "OK",
"detail": "Adresse identique sur 4 pieces"
},
{
"controle": "falsification_pdf",
"statut": "OK",
"score_falsification": 0.008,
"detail": "Aucune couche suspecte"
},
{
"controle": "coherence_revenus",
"statut": "OK",
"detail": "Ecart cumul vs avis: 0.2%"
},
{
"controle": "iban_beneficiaire",
"statut": "OK",
"detail": "IBAN correspond au titulaire"
},
{
"controle": "detection_clones",
"statut": "OK",
"comparaisons": 847291,
"detail": "Aucun clone detecte"
}
],
"temps_traitement_ms": 342
},
"ingestion": {
"systeme_cible": "GED_Alfresco",
"statut": "INGERE",
"dossier_cree": "ALF-2024-08471",
"pieces_indexees": 6,
"metadonnees_injectees": true
}
}
Provence AI

Every document read, every field captured.

Your teams spend endless hours keying, checking and filing information by hand on every case.

Tedious, time-consuming, low-value work.

Our clients automate end-to-end: capture, cross-checks, controls, delivery. That simple.

One platform. Four phases. Seven steps.

From the incoming batch to the data delivered into your IT - everything is integrated. Not three tools, not five vendors.

  1. Phase 01

    Intake

    We split and we classify.

    • Splitting

      Batches → documents

    • Classification

      Type · naming

  2. Phase 02

    Understanding

    We read and we cross-check.

    • Extraction

      Fields, tables

    • Reconciliation

      Reference data, external sources

  3. Phase 03

    Verification

    We check and we hunt fraud.

    • Checks

      100+ business rules

    • Fraud detection

      Document forgeries

  4. Phase 04

    Delivery

    We push into your IT.

    • Delivery

      API, SQL, DMS, ERP

Explore document processingExplore fraud detection

Nous co-construisons avec

Vaucluse
CNES
Ardèche
Vienne
Eure-et-Loir
Loir-et-Cher
Nexpublica
Arche mc2
Worldline
Digitech
Landes
Essonne
Haute-Garonne
Charente-Maritime
Efalia
AtolCD
Stimut
Allianz

Let's talk about your document workflows.

30 minutes. No sales deck, no questionnaire. Your cases, our answers.

Request a demoDocument processingFraud detection