How a major fintech is using GenAI to revolutionize financial document validation
- Marcelo Firpo
- Feb 26
- 4 min read
Automation in document processing Reduction of operational costs Scalability and accuracy with generative AI
AI generated summary:
A large Brazilian fintech company specializing in digital automation faced challenges in validating financial documents, dealing with high volumes of manual processing, high costs, and frequent errors. To solve this problem, BlueMetrics developed a solution based on Generative AI, which automates the extraction, categorization, and processing of identification documents with high accuracy and scalability.
Overview
The client is a fintech with over 20 years of experience in the market, offering innovative solutions in digital automation for processes and documents. The company stands out for its full-service approach, supporting medium and large companies in their digital transformation. Its clients include some of the largest banks in the country.
In a highly competitive and regulatory environment, the client was looking for solutions that would increase its operational efficiency, improve the experience of its own customers and ensure compliance with financial sector regulations.
Market context:
Growing demand for agile and secure digital processes.
Need for compliance with financial regulations.
High competitiveness requiring more efficient processes.
Customer expectations for more fluid digital experiences.
Problem: How to make document processing more efficient?
The traditional OCR solution used by the client had limitations that directly impacted operational efficiency and business growth. According to Gabriel Casara, CGO at BlueMetrics, “We are talking about hundreds of thousands of documents that need to be correctly processed every month. This is an exemplary use case for generative AI and a great opportunity to develop work that actually solves a clear customer pain point.”
Operational limitations:
High volume of manual document processing.
Frequent errors in data extraction and categorization.
Bottlenecks that increased onboarding time.
Business limitations:
Difficulty scaling operations without expanding the backoffice.
High image processing costs.
Compromised customer experience during registration.
Technological limitations:
Limited and poorly adaptable OCR technology.
Need for a modern and scalable solution.
The solution: automation and increased accuracy through GenAI.
BlueMetrics has developed a multimodal Generative AI solution to automate and optimize the processing of identification documents, highlighting the following points:
Automated processing of driver's licenses and IDs in different formats and orientations.
Accurate data extraction such as name, date of birth and document number.
Automatic categorization of the extracted data, reducing manual intervention.
The solution uses Generative AI to automatically correct the orientation of images, extract data with high precision and categorize information, offering agility and reliability. It is mainly applied in the contexts of KYC (Know Your Customer), account opening and identity validation and automation of registrations and contractual processes.
According to Diórgenes Eugênio, Head of Gen AI at BlueMetrics, “The project developed with this client to extract information from identification documents had several phases and approaches, considering that there are several possibilities for extracting organized textual information from images. The first major obstacle we identified was that the orientation of the images directly affected the models’ ability to identify and extract the information. After identifying this obstacle, we explored some solutions, including using Generative AI models to identify the number of degrees needed to leave the image in a standard orientation. However, the results were not satisfactory. Therefore, our final pipeline used Tesseract running alongside an API to identify the need for rotation. After that, we evaluated several techniques used to improve the quality of the images, improving the character recognition capacity. For the specific context of identification documents, not all of them presented significant gains. The project generated a major impact on the business, considering that the major difficulty of OCR solutions today is organizing the information, correlating the extracted text with the information in the document. In addition to this advancement, the architecture we created proposes constant evolution, as we increasingly have cheaper and more capable models.”
Main features
The solution brought some important technological differences, capable of generating positive impacts not only on the effectiveness of processes, but also on the customer experience, thus translating into concrete results for the operation as a whole.
Intelligent preprocessing:
Automatic detection of document orientation.
Correction of image positioning and optimization.
Advanced Extraction Pipeline:
Multimodal Generative AI models for data mining.
Intelligent categorization system and parallel processing.
Solution differentials:
Adaptability: compatible with different documents and guidelines.
Precision: reduced errors and greater accuracy in data extraction.
Scalability: cloud-native architecture to process large volumes of documents.
How about developing a solution like this for your company?
Results:
The implementation of the solution brought a series of benefits to the client, resulting in concrete and immediate financial results.
Operational Efficiency
Reduction of operational costs and average onboarding time;
Elimination of bottlenecks in document processing;
High productivity with greater simultaneous processing capacity.
Quality and precision
High accuracy in data extraction and categorization;
Significant reduction in errors and rework.
Business impacts
Greater scalability to meet peak demand;
Flexibility to process multiple types of documents;
Evolutionary solution, aligned with future needs.
Compliance and security
Full traceability and compliance with financial regulations;
Improved detection of fraud attempts.
Technologies used
The solution was designed using several AWS technologies, including:
AWS Services
ECS
Lambda
Bedrock
S3
DynamoDB
EventBridge
Languages, Libs and Frameworks
Python
Tesseract
OpenCV
Conclusion:
With the new Generative AI solution, the client revolutionized its document validation processes, consolidating its position as a leader in innovation in the financial market. “The partnership with BlueMetrics demonstrated how advanced technology and an agile implementation methodology can transform operational challenges into lasting competitive advantages,” adds Gabriel Casara.
How about creating a case like this for your company? Let's schedule a call?
Discover some Use Cases .
About BlueMetrics
BlueMetrics was founded in 2016 and has already delivered more than 160 successful solutions in the areas of Data & Analytics, GenAI and Machine Learning for more than 70 companies in the United States, Brazil, Argentina, Colombia and Mexico. It has its own methodology and a multidisciplinary team focused on delivering solutions to real challenges in the business world.
Comments