Case Study — DTX Intelligence™ · Innova Black®

The Client

A regulated SOFOM (Sociedad Financiera de Objeto Múltiple) based in Guadalajara, México, with an active credit portfolio serving individual and business borrowers. The institution operates under CNBV supervision and is subject to PLD/KYC compliance requirements under Mexican financial regulation.

The Challenge

The institution’s credit analysts were manually reviewing applicant documentation — bank statements, official ID, proof of address, proof of income, credit bureau reports — and producing written credit assessments from scratch for each file. The process was time-intensive, inconsistent in structure across analysts, and created bottlenecks at the final dictamen stage before committee review.

The goal was not to remove the analyst from the process. The goal was to give them a structured, complete first draft of the credit dictamen at the moment they needed it most — so their judgment could focus on the decision, not the documentation.

The Solution — DTX Intelligence™ Credit Module

Innova Black® deployed Claude as the AI engine at the dictamen generation stage of the credit workflow, integrated within the institution’s existing operations via the DTX Intelligence™ framework.

The flow works as follows:

  1. Document ingestion. Applicant documents are submitted digitally: bank statements, official government ID (INE), proof of address, proof of income, and credit bureau report. Claude reads and processes each document as part of a structured analysis pipeline.
  2. Data extraction and structuring. Claude extracts and organizes the relevant data points from each document into a standardized credit file structure: identity verification, address confirmation, income analysis, payment behavior from bank statements, and credit history summary from bureau data.
  3. Dictamen generation. Claude produces a complete written credit dictamen in institutional Spanish, structured according to the SOFOM’s internal credit policy. The dictamen covers applicant profile, financial capacity assessment, risk observations, and a preliminary recommendation — ready for analyst review.
  4. Human review. The credit analyst receives the structured dictamen, reviews Claude’s analysis against the source documents, applies their judgment, and signs off before the file moves to credit committee.

The Result

The analyst enters the review stage with a complete, consistently structured dictamen rather than a blank document. The time previously spent on documentation is redirected to decision-making and exception analysis. Credit file quality is uniform across analysts and loan types.

The institution maintains full human accountability on every credit decision — Claude accelerates and structures the process, the analyst owns the outcome.

Technology

The integration leverages Claude’s ability to process complex, multi-document inputs in Spanish and produce structured institutional-grade outputs — a capability that distinguishes this approach from generic automation tools. The DTX Intelligence™ framework manages the orchestration layer: document intake, prompt engineering calibrated to the SOFOM’s credit policy, output formatting, and audit trail generation required for regulatory compliance.

For institutions evaluating the application of AI within regulated credit workflows, this case study demonstrates a deployment model where the technology serves the analyst rather than replacing them — and where every output remains subject to human judgment before any credit decision is made.

To learn more about how Anthropic’s Claude is being deployed in Mexican financial services, read our analysis on Anthropic’s technology in fintech.

DTX Intelligence™

Does your institution need to automate its credit process?

Request a free diagnostic. We assess your current credit workflow, identify automation opportunities, and deliver a clear implementation roadmap tailored to your institution’s regulatory requirements.

Request Free Diagnostic