Client name
BNP Paribas Leasing Solutions
Client type
Company from the financing sector
Industry
Finance / Leasing / Sale and leaseback / Property leasing / Loan leasing
Context
Every financial contract (leasing, loan, lease) requires analysis of a set of documents submitted by the client. Before implementing Nu Delta’s docAI technology, this process relied on manual review of various document types—from contracts and declarations to payment confirmations. Due to the volume and diversity of documents, the process was time-consuming, error-prone, and difficult to scale.
Documents in PDF and JPG formats were submitted via the LEO database, with each case containing a dozen or more different documents. Nu Delta’s GDB Tool was implemented as part of the process to automate data extraction and verification—from document type recognition to logical analysis and decision-making.
Effects and results obtained
In the entire process, we prioritize the end user by applying a “Human-in-the-loop” approach. After the automatic data extraction from scanned documents, the system operator receives a processing summary and makes a decision:
If the data is correct, they approve it.
If Nu Delta’s docAI flags missing data, the operator can fill in the gaps.
If docAI indicates erroneous data, the operator can correct it.
The operator can also simply reject the application.
Results:
Automatic document type recognition across diverse inputs (up to 15 document types per client)
Automatic extraction of predefined fields from documents (e.g., PESEL numbers, dates, amounts, tax IDs, address data)
Data verification according to rules set by the administrator
Reduction of human errors in document evaluation
Faster application assessment and increased transparency in the decision-making process
Generation of a summary (PDF) for the user with processing results
Client goals
Automation of data verification in documents required for leasing agreements
Reduction of operational workload for analytical and back-office departments
Standardization and unification of document analysis regardless of file format or source
Shortening the processing time for each application
Starting point
Manual checking and data entry from PDF/JPG documents
Need to train the team to process many different document types and formats
Risk of human errors and inconsistent decisions
Legal and compliance risks
No centralized report summarizing analysis results
Document types covered by the implementation
Client statements, payment proofs, identity documents, contracts, annexes, powers of attorney
Format: PDF, JPG, PNG
Scope and course
Why Nu Delta: The client valued experience in automating documents with diverse structures and ease of configuring business rules.
What we did:
Each document batch (so-called input) related to a single leasing agreement may contain from several to a dozen different document types, such as statements, contracts, attachments, identity documents, or payment confirmations.
The automation process using GDB Tool runs in several stages:
Document type recognition – the system automatically identifies the type of each file submitted by the document database (Nu Delta recognizes if the document is, for example, an ID card, application form, or certificate).
Data extraction – for each recognized document type, predefined fields are read to retrieve and verify data (e.g., PESEL, NIP, amounts, dates, addresses).
Business rule verification – data from documents and accompanying data (e.g., from the client form) are automatically compared and evaluated according to rules defined by the administrator, such as data consistency, document completeness, required fields, or value ranges.
Results presentation – the end user (e.g., analyst or operator) receives a processing summary and makes a decision:
If data is correct, they approve it, or if:
Nu Delta indicates missing data, they can add it,
Nu Delta indicates erroneous data, they can correct it,
The operator can also reject the application.
Report generation – the system automatically creates a PDF file with a full summary of extraction and verification results.
Technology: docAI platform by Nu Delta (OCR + NLP + business rules engine), operating in on-premise or cloud model.
Security: Full compliance with GDPR and client IT standards.
Integrations and how it works
Integration with the LEO document database – documents automatically transferred to GDB Tool.
Transfer of processing results to the client’s decision-making system.
User reporting with a summary of the analysis.
Implementing GDB Tool allowed us to automate document analysis in leasing and loan processes. The system independently reads data, identifies inconsistencies, and significantly speeds up application assessment, while enhancing the security of decisions from a formal and legal standpoint.
