CONFIDENTIAL CLIENT
How a Brazilian IT company is leading the digitization of historical archives for universities
Digitize and make university historical archives searchable
Innovation in digitizing historical archives. Automation in research and cataloging. Scale through artificial intelligence.
A Brazilian company that specializes in digital automation and document management worked with BlueMetrics to build an AI solution that modernizes access to the historical archives of higher education institutions. The project automates how historical documents are extracted, organized, and searched, using semantic search and image processing to structure information in context.
Overview
The client is a technology company with more than 20 years in the market, offering digital automation solutions for processes and documents. It is a reference in digital automation and document management in Brazil, known for helping large higher education institutions with their digital transformation.
The Smart Search for Newspaper Archives project was created to meet growing demand for digitization and efficient access to historical information. It speaks directly to challenges faced by libraries, public archives, universities, and media organizations.
Market context
- Growing demand for digitizing and organizing historical archives.
- The need to preserve unique, valuable documents.
- The push for faster, more accurate document search.
The problem: how do you improve the research experience in historical archives?
Searching historical archives is genuinely hard, and it affects operational efficiency, information quality, and an organization’s room to grow. The main obstacles include physical documents that are deteriorating, outdated technology, and the difficulty of returning accurate, in-context results. The result is slow processes, high costs, and a user experience that falls short, all of which cap how far these services can scale and compete.
As the team put it, “This is one more practical example where AI can genuinely make a difference day to day, speeding up processes and freeing teams for more strategic work and less manual labor.”
Main challenges:
- Operational
- Technological
- Business
The solution: automation and scale with AI
Starting from that need, BlueMetrics built a solution that combines several technologies to modernize access to historical archives.
According to Diórgenes Eugênio, Head of Gen AI at BlueMetrics, “This was easily one of the most demanding projects of the year. On top of the complexity of working with deteriorated source material, we had to organize the text while keeping the relationship between an article’s headline and its body intact. That was the real challenge, because extraction is unstructured: each word is pulled out with no relationship to the others. To solve it, we tried a few strategies, such as using the coordinates of the extracted words to reassemble the text in a logical order. We also used the sizes of the identified text boxes to separate article text from headline text. That last approach made a real difference in how well the large language models could reconstruct those relationships. Those were the main challenges of the first component, extracting the information. Once we got past that stage, we ran into the challenges of the second component, search. In search, the big obstacle was making sure every article had a genuinely relevant semantic match. For that, we drew on the literature for a few approaches, mainly using confidence scores on the returned vectors.”
Key features
- Context preservation: the historical and documentary context is kept intact.
- Advanced semantic search: more accurate, more relevant results.
- Process automation: less search time and more efficiency.
- Digital scale: infrastructure built for large data volumes.
Technology components:
- Smart extraction system
- Semantic search engine
- Technical innovations
Results
The solution delivered meaningful gains across operational efficiency, information quality, and commercial strategy. With the right technology in place, the client streamlined its search processes, preserved the historical context of the data, and expanded how far its operations could scale. These results turned the original challenges into competitive advantages and reinforced the value of modern access to historical archives.
“This solution has enormous potential to solve similar problems in other kinds of companies and industries, and it fits easily within our proprietary way of working, the blue4AI method.”
Operational benefits:
- Up to an 80% reduction in document search time.
- Automated extraction of structured data.
- A significant increase in how many requests can be handled at once.
Technological benefits:
- Modern access to archives.
- Scalable infrastructure for large data volumes.
- Historical documents preserved in standardized formats.
Strategic benefits:
- A competitive edge in the market.
- Room for new business models and partnerships.
- A better experience for the end user.
Technologies used
The solution was built on a range of AWS technologies, including:
AWS services
- Textract
- Lambda
- Bedrock
- S3
- DynamoDB
Languages, libraries, and frameworks
- Python
- Pillow
- Fitz
- FPDF
Conclusion
Thanks to this solution, the client cleared major hurdles in digitizing and searching archives, and it further cemented its position as a leader in digital transformation by giving its higher education clients a faster, more accurate, and more scalable research experience. By combining automation, artificial intelligence, and context preservation, the company changed how historical information is accessed and used.
This model benefits schools, public archives, and libraries, and it opens the door for other organizations, public and private, across any industry, to put AI to work improving their own processes and services.