The platform

Document technology. Specialised for your workflow.

MestrIA is the document intelligence platform that brings together document reading, classification, extraction and cross-checking. Its solutions include proprietary models, fields and rules for debt collection and real estate assets.

Explore document types →
Documents enter the processing flow and are organised for review
Documents → Classification, extraction and reconciliation → Data and issues
▪ Input and output ▪

Input documents. Information ready for your system.

How documents come in

Upload batches of files or connect an agreed mailbox, SFTP, API or shared drive. For email workflows, MestrIA reads the message and attachments, identifies the case and extracts the agreed fields. Access and processing rules are configured for the project.

How the data comes out

Receive results in Excel, CSV or JSON, or in your system’s format. Each field includes the value, confidence score and source reference: document, page and position.

▪ The process ▪

Four steps from documents to review

MestrIA organises documents by case file, extracts the agreed fields and cross-checks the data against other sources. The results identify discrepancies and missing documents for your team to review.

01

Separate and classify

MestrIA separates documents within a file, recognises their type from their content and associates them with a case or property using the agreed identifiers. Unreadable or ambiguous documents go to review.

02

Extract field by field

Extracts parties, amounts, financial terms, collateral, assignments and signatures. Each value is linked to its source page and confidence level.

03

Cross-check and reconcile

Compares parties and amounts across contracts, deeds and inventories. Checks received documents against those declared by the seller.

04

Prepare the review

Identifies complete files, discrepancies and missing documents. The team can review the details and prepare subsequent actions.

Checks are adapted to the stage of the transaction, from initial analysis to post-acquisition management.

Classification in motion

Demo · Simulated data

In this simulated example, a batch of 12,480 files arrives mixed and with useless filenames. The platform identifies each document by type and places it in its loan file: agreement, balance certificate, transaction statement and signature evidence. Anything that cannot be classified goes to an exceptions queue. With the file complete, the data is cross-checked between documents (borrower, balance and signature) and the file is prepared for professional review.

Batch of 12,480 files · 4 documents per file · 2 to exceptions

File MC-3744408 Martínez Solé, E.
Agreement personal loan · 9 pp. ✓
Balance certificate as at 30/06/2024 · €4,812.44 ✓
Transaction statement 04/2019 to 06/2024 · 63 entries ✓
Signature evidence qualified seal · 12/03/2019 ✓
Cross-checks
✓ borrower · agreement = certificate Martínez Solé, Elena
✓ certified balance = sum of transactions €4,812.44
✓ signature · seal and timestamp recorded 12/03/2019
File complete Ready for professional review 4 of 4 documents · 3 cross-checks · traceable per field

1,847 files ready · 293 waiting on a document

▪ Models and processes ▪

Fields and checks for each document type

Specialisation defines which documents to identify, which fields to extract and which relationships to check. Debt certificates are compared with agreements; asset documentation is reviewed for type, validity dates and consistency.

Fields and checks for each type

The project scope specifies the fields to extract, the checks to perform and the evidence to retain.

Confidence field by field

Each value includes a confidence score. Values below the agreed threshold are flagged for human review.

Checks against the source text

MestrIA checks extracted values against the source text. Values without support in that document are excluded from the output.

Accuracy is assessed by field and document type using a representative sample. Targets and the handling of lower-confidence fields are agreed for the project.

Field-level source evidence

Demo · Simulated data

Personal loan agreement p. 1 / 9
Borrower MARTÍNEZ SOLÉ, ELENA
NIF 47829153K · address C/ Marina 142, 4º 2ª, Barcelona
Signature date 12 March 2019

Principal advanced €7,490.00
Term 60 monthly payments · instalment €158.42
Nominal rate 8.25% · APR 9.14%

Clause 10.2 · default interest
Two points above the ordinary rate

Co-borrower FERRER OLIVA, ANDRÉS
MestrIA Engine v3 evidence per field
titular0.99 Martínez Solé, Elena
nif1.00 47829153K
capital_inicial0.99 €7,490.00
tae0.98 9.14%
cotitularreview Ferrer Oliva, A. · not in the data tape
5 of 34 fields p. 1 · entry 4
▪ Modules ▪

Document solutions by use case

Solutions are combined to suit the documents and operational requirements.

01

OAP reading and classification

Reads PNJ reports, identifies balances, income and assets, and automatically classifies results by debtor. Prioritise cases with identified assets and distinguish no findings from cases needing more information.

02

Land registry extracts

Converts land registry extracts into structured property, ownership and charge data. Supports asset analysis and flags ambiguities and normalisation issues.

03

Document preparation for legal claims

Brings together data from agreements, certificates and other case documents. Cross-checks information and flags inconsistencies and missing documentation before legal review.

04

Portfolio document due diligence

Cross-checks portfolio data against the available documentation. Identifies differences, information without documentary support and missing documents to guide the review of each transaction.

05

Bulk contract extraction

Transforms batches of agreements into structured data on parties, amounts, dates and financial terms. Organises results for analysis, review and use in your processes.

06

NPL portfolio document boarding

Prepares the document inventory and data needed to bring a portfolio into your operations. Organises documentation by transaction and defines checks and delivery format according to your management system.

07

Judicial case reconstruction

Classifies documentation and orders the milestones needed to reconstruct the history of proceedings. Links each milestone to its source and makes gaps in the available documentation visible.

08

Court notice classification

Identifies the type of decision, court, proceedings, parties and time limits stated in each notice. Prepares the information for review and transfer to your daily operations.

Real Estate

Document management and auditing for real estate assets

Classify registry extracts, certificates, permits and reports by property. Review validity dates, consistency and missing documents; receive an asset record and issues. The checks and delivery format are configured for your workflow.

See documents and output →
Intake and integrations

From the inbox to a reviewable case file.

Documents can arrive individually, in a portfolio batch or attached to an email. MestrIA identifies the case, structures its information and prepares the agreed output for your management system.

01 · Email and attachments

Keep the message and its documents together.

Reads the message and attachments from the configured inbox. Extracts references to the proceedings, identifies the case and prepares structured data for review and delivery.

Output: associated case, extracted fields and uncertain matches for review.

02 · Portfolios and batches

One inventory across thousands of documents.

Splits and classifies the supplied files, groups them by case or asset, and checks data and completeness against the agreed inventory.

Output: document inventory, structured records, discrepancies and missing items with source evidence.

Available channels

Batch upload · Email · SFTP · API · Shared drive

Agreed output

Excel · CSV · JSON · Your management system

Connections, permissions, matching rules and field mapping are defined for your project. A mapping for Kmaleon, Tallyman or your own system depends on its available interfaces and the agreed scope. Ambiguous matches remain available for professional review.

▪ Deployment ▪

A managed EU platform and options defined by contract

Managed in the European Union

The main offering is the MestrIA platform managed in the European Union.

In your cloud

Deployment in the client’s cloud as a project option.

On-premise

Deployment on the client’s servers or on premises as a project option.

Technical scope, data flows, residency, access, processing and responsibilities are defined in the proposal and contract for each deployment.

▪ Data processing ▪

Data, permissions and retention

The proposal and contract specify where data is processed, access permissions, retention and return.

Data protection

The contract sets out the project’s security measures and confidentiality commitments.

Permissions and activity records

The agreed scope defines access profiles, permissions and activity records.

References to ISO/IEC 27001, ISO/IEC 27018 and SOC 2 relate only to infrastructure providers and the services covered by their certificates or reports. They are not MestrIA certifications. Applicable documentation is identified for each project.

For operations under the Directive

Directive (EU) 2021/2167 regulates credit servicers and credit purchasers. The platform produces the per-field traceability and the access log that those operations have to document.

The technology behind your solutions

Advanced engines. Our own sector intelligence.

We combine OCR and extraction technologies with proprietary models trained for specific sector document types. Each solution includes its classification, fields and checks.

How the platform worksIllustrative example · Fictional data

01 Your documents

PDF / 01Loan agreementA. García
Case file 0042
PDF / 02Account statementBalance: €3,200.00
30 Sep 2026

Folders, PDFs and scans. Every document retains its source.

02 Document reading

Illustrative route
Mistral OCR
Docling
Text · Tables · Structure

One or more engines, depending on the document and deployment.

03 Sector intelligence

Criteria for each solution

Document types · Fields to extract · Agreed requirements · Reference datatape

MestrIA

Document engineering and processes

  1. Classification

    Identifies each document and assigns it to its case file.

  2. Extraction

    Captures borrowers, products, dates and amounts.

  3. Audit

    Cross-checks documents, data and requirements; flags exceptions.

Our own specialised models, such as MestrIA-v3-2B.

04 Your result

Structured data / 0042
Borrower
A. García
Product
Consumer loan
Balance
€3,200.00
Balance discrepancy

Datatape: €3,150.00.
Same field and date: 30 Sep 2026.

Source: statement · p. 2
Excel · JSON · API

Integrated engines

The route is configured for each solution; not every engine is used on every document.

  • Mistral OCR
  • LlamaParse
  • Amazon Textract
  • Azure AI Document Intelligence
  • Docling
  • MinerU
  • Google Document AI
European Union

Data and processing in the EU

Deployment options on your own servers.

Access, retention and deployment conditions defined for each project.

  • Controlled access
  • Document traceability
  • Contractual confidentiality

Set up the solution for your team

We start with the solution for your workflow and check its fit against a document sample. Coverage, outputs and connections are agreed; any adaptations are defined before onboarding.