REQUEST: The company’s team was looking for a reliable partner with expertise in artificial intelligence and cloud infrastructure to automate work with archives and current documentation. The main task was to implement an optical character recognition (OCR) system that would allow for the intelligent systematization of document scans, minimizing manual labor and the likelihood of errors.
RESULT: Wise IT specialists developed a desktop application based on Gemini 1.5 Pro, which ensured accurate data recognition and its integration into a local database, fully satisfying the customer’s security requirements.
Business Challenge
The main challenge of the project was the need for uncompromising protection of strategically important data. Due to the status of a state-owned enterprise and the specific nature of the industry, the solution had to comply with strict regulatory security requirements. It was necessary to create a system that combines the efficiency of cloud OCR mechanisms with the secure processing of the customer’s sensitive information.
Solution
The Wise IT team developed a custom desktop application that utilizes the capabilities of Gemini 1.5 Pro LLM via Google Cloud Platform, which was tailored by the enterprise’s specialists taking into account information security requirements.
Features of the implemented solution:
- Intelligent OCR: The application is capable of recognizing complex structures of scanned documents, regardless of their type.
- Data Extraction: Artificial intelligence analyzes the content, extracts key information, and automatically structures the documents.
- Local Integration: Processed information is automatically entered into the customer’s local database, ensuring confidentiality and control over data.
- Security: The developed architecture provides reliable data protection and does not store it outside the customer’s local infrastructure.
Result
The enterprise team received a modern tool that:
- Significantly accelerated the processing of archival design documentation.
- Ensured high recognition accuracy thanks to the use of the multimodal Gemini AI model.
- Allowed the team to use Google Cloud tools in a secure and efficient manner.
Gemini 1.5 Pro and OCR Technologies
Gemini 1.5 Pro is a multimodal LLM model from Google that features an expanded context window. It is capable of processing large volumes of information, understanding not only text but also complex tables, charts, and logical relationships within documents.
Optical Character Recognition (OCR) is a fundamental technology that converts images (scans, photos) into text understandable by a computer. Combined with large language models (LLMs), OCR evolves from simple character reading to intelligent analysis, enabling automatic document classification and meaning extraction.
Customer Feedback
“The pilot project based on cloud OCR with handwritten text recognition in design documentation made it possible to digitize the archive of design documentation within an acceptable timeframe, which is particularly important given the difficult circumstances the enterprise faces during the Russian-Ukrainian war.
The experience of the Wise IT team, their expertise, and their active role as advocates for emerging technologies made it possible to consider and evaluate the feasibility of such a project and to shape our own permanent solution.
It is important to note the diligent work and the Wise IT team’s understanding of the state-owned enterprise’s internal environment and the system of limitations regarding information security and data protection, which allowed for efficient and fruitful cooperation in implementing new tools.”
— Official representative of the enterprise.
