Automating Document Recognition with Gemini AI for NovaPay

Automating Document Recognition with Gemini AI for NovaPay

NovaPayis an international financial service from the NOVA group of companies (Nova Poshta), which has been creating convenient financial solutions for individuals and businesses for 14 years, operating online and offline in more than 3,600 branches across Ukraine. The NovaPay app is an innovative financial platform with separate profiles for personal and business needs. NovaPay processes millions of financial transactions daily, ensuring high speed, accessibility, and reliability of payment services for millions of customers.

REQUEST: The NovaPay team reached out to Wise IT with a request to build a solution for an automated customer verification process using artificial intelligence technologies.

RESULT: The joint team of NovaPay and Wise IT developed and tested an architecture based on Google Cloud and Gemini AI, which reduced document processing time from 1–2 minutes to 3–5 seconds with an initial recognition accuracy of 85% and ensured complete data security.

Business Challenge

As part of its digital transformation strategy and service quality improvement, NovaPay worked on optimizing the customer identification process during financial transfers or parcel collection at branches. The company was looking for a solution to transition from traditional manual document verification methods to a proactive and fully automated intelligent document processing (IDP) system.

One of the key stages of the identification process was the manual entry of passport data (booklet or ID card) and Taxpayer Identification Number (TIN) by operators. The process took 1 to 2 minutes, which during peak hours could increase customer service time.

Additionally, the process was complicated by “noisy” data factors:

  • Documents were often scanned or photographed at an angle, with glare from glossy surfaces or under low light conditions at the branch.
  • Old-style booklet passports contained illegible handwritten text (for example, in registration stamps or the “Issued by” field).

Any typos by operators created serious legal risks. At the same time, the fintech structure had to ensure absolute confidentiality and protection of financial secrets in compliance with strict corporate standards.

Solution

The joint team of NovaPay and Wise IT developed a solution built on the Edge-to-Cloud (branch ➔ cloud) concept in a secure Google Cloud Enterprise environment. At the core of the architecture lies the synergy of two tools: the open-source OpenCV library and the Gemini multimodal model. Wise IT specialists proposed image preprocessing based on OpenCV; the NovaPay team independently integrated the Gemini model via Vertex AI for intelligent data extraction and validation.

The processing and validation process was automated in a few steps:

  • Primary Processing via OpenCV: When an operator takes a quick photo or scan of a document, OpenCV automatically splits the image into separate parts and individually removes noise and glare for each single element.
  • Intelligent Control with Gemini: The neural network acts as an intelligent extractor and validator, performing cross-verification of data. Thanks to optimization for Ukrainian documents, the model clearly recognizes their structure and instantly parses text fields. With its powerful contextual vision, Gemini reads even illegible handwritten text from old stamps.
  • Enterprise Security: All operations take place inside an isolated NovaPay cloud environment. Customers’ personal data is encrypted, not shared with third parties, and not used by Google to train public models.

Result

Pilot testing of the AI system demonstrated a positive economic and operational effect:

  • Significant Speedup (AHT): The time for entering and verifying customer data was reduced from 1–2 minutes to 3–5 seconds.
  • High Accuracy: Already at the first stages of recognizing real “noisy” documents, the system achieved 85% accuracy.
  • Minimization of the Human Factor (Zero Errors): The risk of entry errors in financial forms has been minimized.
  • Scalability and Easy Start: The system has proven its readiness to quickly process documents on a large scale. The next stage involves expanding the project to 5,000 documents, with further scaling to a stream of 1M+ documents from thousands of branches across Ukraine.

Tech Stack

Gemini is Google’s flagship multimodal artificial intelligence model. Its uniqueness lies in its ability to simultaneously understand, combine, and analyze different types of information: text, images, audio, and code. It allows businesses to reach the level of cognitive context analysis and real-time recognition of complex handwritten fonts.

OpenCV is a leading open-source computer vision library. It is used for rapid document image preprocessing: automatic alignment, noise and glare elimination, and precise structure recognition (pages, ID cards) for further analysis.

Customer Feedback

“For NovaPay, speed of service is not just about technology, but also about customer comfort. Every day we process millions of financial transactions, so even a few dozen seconds saved during identification have a tangible effect across the entire network. That is why we are constantly looking for opportunities to automate routine processes and implement modern AI solutions.

In partnership with Wise IT, we managed to build a solution that automated document recognition and significantly reduced processing times. For customers, this means faster service, and for our teams, fewer manual operations and more time to focus on the customer. At the same time, it was of fundamental importance to us that the new solution met the highest security and personal data protection requirements. This project was another step in NovaPay’s digital transformation and the development of services that make financial services simpler, faster, and more convenient.”

– Yuriy Kasyan, Project Manager for AI Implementation, NovaPay.

Other Clients
Gemini OCR for Document Systematization at a Ukrainian State Enterprise
Gemini OCR for Document Systematization at a Ukrainian State Enterprise
The team of a Ukrainian state enterprise required an innovative, AI-powered OCR solution to systematize large volumes of scanned documents while maintaining the highest standards of information security.
Read a success story
Google Cloud infrastructure for Finhike
Google Cloud infrastructure for Finhike
Finhike required a flexible and secure cloud infrastructure to scale its fintech services globally and ensure high availability even with unstable connections. By migrating to Google Cloud, the company achieved a 99.95% uptime rate, accelerated engineering tasks by 20% using Gemini, and reduced annual infrastructure costs by 10%.
Read a success story
Configuring Google Workspace  according to international information  security standards for Crimson IT
Configuring Google Workspace according to international information security standards for Crimson IT
Crimson IT needed help implementing robust information security measures for its Google Workspace environment. The company turned to the Wise IT team with this request.
Read a success story
Teaching the marketplace to understand the customer: implementing GenAI for natural language product search for VARUS.UA
Teaching the marketplace to understand the customer: implementing GenAI for natural language product search for VARUS.UA
The VARUS.UA team approached Wise IT with a request to implement intelligent product search for users of the VARUS online marketplace that would allow the use of natural language.
Read a success story
PROSTOR team migration to Google Workspace
PROSTOR team migration to Google Workspace
The PROSTOR team turned to Wise IT to migrate the team from an outdated email service to the Google Workspace platform.
Read a success story
Corporate knowledge base with Google Sites and Vertex AI for NOVA Group
Corporate knowledge base with Google Sites and Vertex AI for NOVA Group
The NOVA Group of Companies turned to Wise IT with the need to create a centralized corporate resource — a knowledge base for company employees with intelligent search for the necessary information.
Read a success story
Gemini AI for the VARUS team
Gemini AI for the VARUS team
The implementation of Google's AI-based tools, implemented together with the Wise IT team, allowed VARUS to transform daily work processes, shifting employees' focus from routine tasks to analytics, creativity, and strategic decision-making.
Read a success story
Optimizing Google Workspace for Bob W
Optimizing Google Workspace for Bob W
Bob W team was looking for a more flexible and effective Google Workspace licensing solution that would accurately reflect the diverse roles and needs of their employees, moving away from their existing one-size-fits-all model.
Read a success story
Google Workspace for Ivmed
Google Workspace for Ivmed
Google Workspace for the Fertility Health Center Ivmed IVMED is a fertility health center for couples. The company helps couples consciously approach the birth of a child through healthy conception, from the moment they decide to become parents throughout the entire pregnancy period. Healthy conception includes both natural conception and IVF at IVMED. With IVMED, infertility does not exist; there is only a way to achieve the desired pregnancy with the help of modern fertility technologies.
Read a success story
Google Workspace for Preply
Google Workspace for Preply
Preply has been successfully using the standard Google Workspace plan for several years, but their rapidly growing business requires unlimited access to shared office tools and a higher level of efficiency in corporate communications.
Read a success story
Google Workspace for Energy Efficiency Fund
Google Workspace for Energy Efficiency Fund
Thanks to the flexible and simple tools of Google Workspace, the foundation was able to create its own website without involving contractors, an IT department, and additional costs.
Read a success story
Google Workspace for Alan
Google Workspace for Alan
Specialists of the Wise IT company help the “ALAN” meat processing plant to evaluate the benefits of Google Workspace services for effective corporate communications.
Read a success story
Google Cloud and Workspace for Tranzzo
Google Cloud and Workspace for Tranzzo
Since 2017, Tranzzo has been creating solutions and technologies that help businesses around the world accept online payments using bank cards. The service conducts about 8 million transactions per month and serves more than 3,000 clients. Tranzzo’s clients include online stores and marketplaces, delivery services, ticket operators, and online schools. In particular, the company cooperates with “Rozetka”, “Arsenal Strahuvaniya”, Prom.ua, WOG, Foxtrot and others. In addition to card payments, the platform works with such payment tools as Google Pay, Apple Pay, Masterpass, VisaCheckout, Telegram Pay.
Read a success story
Google Workspace for Prozorro.Sale
Google Workspace for Prozorro.Sale
Speaking about why the Enterprise level subscription was chosen, Hryhoriy Legenchenko, head of the information technology and telecommunications department, answers: “SOE employees are actively involved in the rule-making processes of Ukraine, we work with a huge number of documents and their versions. Therefore, the unlimited volume of document storage with end-to-end contextual search, which does not require additional configuration, is a determining factor in the effectiveness of our work. A pleasant moment was that for our project we received special conditions for a 24-month subscription and we express our gratitude to the Wise IT company for comprehensive support in any matter”.
Read a success story

Get a free consultation

Fill out the form and our manager will contact you

This site is registered on wpml.org as a development site. Switch to a production site key to remove this banner.