TBC Uzbekistan AI Platform & AI Transformation
- Customer
- TBC Uzbekistan
- Project manager on the customer side
-
- Year of project completion
- 2025
- Project timeline
- January, 2025 - December, 2025
- Project scope
- 45000 man-hours
- Goals
-
- Build an internal AI foundation independent of external vendors.
- Integrate AI into all major business processes and digital products.
- Reduce operational costs through automation and intelligent decision-making.
- Enhance customer experience (CX 360) via personalization and faster service.
- Develop a data-driven and AI-first culture across the organization.
- Project Results
-
- Launched the first AI Assistant for customer interactions in voice and chat channels.
- Delivered two AI-powered products — Sales Assistant and Collections Assistant — into production.
- Reduced the cost of customer interactions by over 90% (from $0.35 to $0.03 per call).
- Achieved a financial effect of $2.3M savings, with a projected $3.7M by the end of 2025.
- Processed over 1.6 million calls and 690,000 chats using AI systems.
- Created a dedicated in-house AI team and ML Competence Center.
- Built the first Uzbek-language LLM (ASR + TTS) and internal data quality benchmarks.
- Established data governance, AI infrastructure, and monitoring pipelines to ensure reliability and compliance.
The uniqueness of the project
- The first full-scale AI Platform developed within an Uzbek bank.
- Combines AI product ecosystem + platform + cultural transformation.
- Designed not only for internal use, but also as a potential AI service provider for the entire TBC holding.
- Includes the AI-ization Program — turning every employee into an active AI user through education and adoption.
- Aligns with TBC’s long-term vision of AI-sation of the enterprise — transforming the bank into a truly AI-driven organization.
- Used software
-
- Machine Learning & AI Frameworks: PyTorch, TensorFlow, LangChain
- LLM Technologies: GPT, open-source LLMs (Mistral, Llama), custom Uzbek LLM
- Data Infrastructure: Airflow, Spark, PostgreSQL, Kafka
- MLOps / DevOps: MLflow, Kubernetes, Docker
- AI Assistant Stack: ASR, TTS, RAG (Retrieval-Augmented Generation)
- Monitoring & Governance: Prometheus, Grafana, internal AI governance tools
- Difficulty of implementation
-
The project required building a full end-to-end AI infrastructure from scratch — including data governance, pipelines, MLOps setup, and internal expertise development — all within a short timeframe.
Main challenges included:
- Lack of local AI expertise and industry experience.
- Complex integrations with existing systems.
- Balancing innovation speed with regulatory and data compliance.
- Shifting mindset and ensuring cultural adoption of AI across teams.
- Project Description
-
The TBC Uzbekistan AI Platform Program is a long-term strategic initiative designed to transform the bank into a fully AI-driven organization.
The program unites technology, culture, and process transformation — enabling every business unit to create, deploy, and use AI-based products. It includes:
- Infrastructure and governance setup,
- A growing portfolio of AI products,
- And a parallel AI-ization track, designed to train employees and integrate AI into daily workflows.
- Project geography
-
Uzbekistan (Headquarters: Tashkent)
- Regional collaboration with companies within the TBC Uzbekistan.
- Local data and AI infrastructure fully developed and hosted within Uzbekistan.