Elevate Customer Trust with AI to a New
Level!

An AI solution for operators that analyzes customer profiles via real-time face recognition and recommends suitable banking services.

See Solution

Problem

Request-Limited Operator

Bank operators are often limited to fulfilling the service the customer came for: issuing a card, making a deposit, or withdrawing money. Meanwhile, other services that could be beneficial to the customer — such as opening cards, high-interest deposits, or special loans — are not offered.

Most importantly, they lack context. They don't know if this customer is a VIP, wants a loan, or recently bought a house. Personalized service and additional sales opportunities are lost.

Result:
• Opportunities to sell banking products remain unused;
• Personalized service to the customer is limited;
• Trust and satisfaction levels between the bank and the customer decrease.

?

Unknown Customer

No data. No personalized offer.

Solution

👁️

Face Recognition

Our camera system recognizes the customer as soon as they enter. If new, it creates a profile. If existing, it retrieves data.

📊

Deep Analysis

Instantly view debit turnover, credit history, and expenses, understanding financial context: traveler, homeowner, or student.

💡

AI Recommendations

AI suggests the perfect card or service based on real needs. "Travel often? Here is our Visa Infinite card."

Result: faster service, fewer errors, increased sales, and a more personalized customer experience.

Why Us?

Harmony of Experience, Knowledge, and
Technology

Our team is a harmonious combination of Data Analysts who deeply understand internal banking processes, experienced developers in artificial intelligence technologies, and professionals who understand real business pains.

99.8% Recognition Accuracy
< 2s Processing Time
24/7 System Uptime
✓

Logical Brain of Artificial Intelligence

Our AI specialist is an expert who not only applies AI but deeply understands its internal logic and mathematical foundations. Alongside Machine Learning and Deep Learning, they perfectly know higher mathematics, statistics, and algorithmic approaches. A champion in algorithms and data structures, with experience in solving complex problems quickly, accurately, and optimally.**

✓

Sufficient Experience in Banking System

Our team leader has real experience working in the banking system and deeply knows card services, limitations faced by operators, and customer communication processes. Also, data analysts in our team have experience working on banking projects.

✓

Strong Technical Competence

Our IT part consists of specialists who have done real projects in AI and Machine Learning, Face Recognition, Data Engineering, Data Analytics, and System Integration. They have high expertise in processing large amounts of data, categorizing transactions, analyzing customer behavior, and creating personal recommendations.

Our Roadmap

Idea

The lack of cross-selling in banks gave us the idea to create an AI platform for personalized customer recommendations.

Nov 2025

Prototype

We developed the initial prototype of the AI recommendation platform. The prototype included the following key functions:

  • Automatic Customer Profiling: Segmentation based on expenses, transaction patterns, banking services, and existing cards.
  • AI Recommendation Generator: Suggests targeted credit, card, deposit, and cashback packages based on customer needs.
  • Simplified UI: Minimal, intuitive interface delivering recommendations to operators in 3–5 seconds.
  • Mini ChatBot Panel: AI assistant answering questions about customer data, spending habits, and services in real-time.

This prototype validated the viability of the core technology, idea, and user experience.

Current Stage Q1 2026

MVP Pilot

At this stage, we have created a Minimum Viable Product (MVP) ready for real-world operation based on the prototype. This version is fully functional and stable for bank operators:

  • Face Recognition (Face ID): Instantly opens profile if customer is found, automatically creates new profile if not.
  • Customer Cards & Balances: All cards, tariffs, and current balances in one place.
  • Service & Transaction Analysis: Debit, credit, deposit, savings, transaction categories (travel, transport, home, health, food, etc.).
  • AI Recommendations: Recommends most suitable cards and products, displaying benefits and tariffs clearly.
  • ChatBot Panel: Understands customer data and answers operator questions in real-time.
  • Fast & Convenient Interface: Opens customer profile in 2–3 seconds, saving time for operator and customer.

Result: MVP is ready for real-world testing in bank branches, improving service quality and creating cross-selling opportunities.

We are here!

Technical Architecture

01

Identification & Data Collection

  • Face Recognition: Customer face detection via face_recognition module.
  • Backend: FastAPI → PINFL / Client Code retrieval.
  • Database: Transactions, cards, credit/deposit history fetched from SQL Server.
  • Frontend: Jinja2Templates (HTML + JS).
02

Feature Engineering & Real-Time Data

  • Generating static/dynamic features from transaction logs (last 30/90 days expenses, P2P networks, liquidity, segments).
  • Online Feature Store: Redis
  • Offline Processing: Spark / PySpark
03

AI Recommendation Engine

  • Hybrid Model: Two-Tower Deep Learning — finding suitable card candidates.
  • Rule Engine: Filtering by age, income, limits.
  • XGBoost Ranking: Ranking the most optimal cards.
  • Explainability (XAI): SHAP → Generating explanations for "Why this specific product?".
04

Chatbot (AI Assistant)

  • Orchestration with LangChain + LangGraph.
  • Ollama GPT-OSS-20B model.
  • Providing accurate answers to operator queries based on pre-prepared SQL/analytics queries.
05

Integration & UI

  • Backend (FastAPI) + Frontend (Jinja2) in a single project.
  • Instant workflow: Face → Identification → Real-time Recommendation → Output with Explanation.
06

MLOps & Deployment

  • Docker / Kubernetes.
  • Model versioning, data-drift monitoring (Prometheus + Grafana).
  • Automated training & deployment via CI/CD.