TBC Bank Uzbekistan completed a comprehensive AI transformation programme in 2025, establishing what is recognised as the first full-scale artificial intelligence platform built and operated within an Uzbek bank. Executed across twelve months and spanning 45,000 man-hours of development effort, the initiative has reshaped the bank's approach to customer service, operational efficiency and technology infrastructure — delivering measurable financial results while laying the groundwork for AI-driven innovation across the entire TBC holding. 

The programme was conceived not as a point solution but as a structural transformation: integrating AI across every major business process, building an internal platform capable of operating independently of external vendors, and establishing a cultural shift that positions AI as a daily working tool for employees at every level of the organisation. The scale and ambition of the initiative set it apart from conventional technology deployments and mark TBC Uzbekistan as one of the most advanced AI adopters among digital banks operating in emerging markets. 

Infrastructure Built from the Ground Up to Operate Independently 

At the foundation of the programme lies a purpose-built AI infrastructure designed entirely in-house. The technical stack encompasses machine learning frameworks including PyTorch and TensorFlow, large language model technologies ranging from GPT and open-source models such as Mistral and Llama to a custom Uzbek-language LLM developed internally, and a data infrastructure built on Airflow, Spark, PostgreSQL and Kafka. MLOps workflows are managed through MLflow and Kubernetes, while monitoring and governance pipelines utilise Prometheus, Grafana and proprietary internal tools developed specifically for TBC's compliance requirements. 

The decision to build this infrastructure in-house rather than relying on third-party AI providers reflects TBC Uzbekistan's broader technology philosophy. By maintaining full ownership of its AI stack, the bank retains control over data governance, model performance and operational reliability — advantages that become increasingly significant as AI systems are embedded deeper into customer-facing and revenue-generating processes. The ML Competence Center established during the programme provides the internal expertise required to sustain and evolve this infrastructure over the long term. 

AI Products Enter Production Across Customer Service and Collections 

The programme's most tangible output is the deployment of AI-powered products into live production environments. The first AI Assistant for customer interactions was launched across both voice and chat channels, handling queries at a scale that would be operationally prohibitive under a conventional staffing model. Two additional AI products — a Sales Assistant and a Collections Assistant — were subsequently delivered into production, each addressing distinct commercial functions within the bank's operations. 

The operational results achieved by these systems are significant. Across 2025, AI systems processed over 1.6 million customer calls and 690,000 chat interactions — volumes that underscore the maturity and reliability of the platform in handling real-world customer demand. Critically, the deployment was not limited to straightforward query routing; the AI systems were designed to manage complex customer journeys across voice and text, maintaining quality standards that support rather than undermine customer experience outcomes. 

Financial Impact Validates Investment at Scale 

[7/20/2026 3:07 PM] SEO Guru: The commercial case for TBC Uzbekistan's AI investment is well supported by the results delivered in 2025. The cost per customer interaction fell from $0.35 to $0.03 following the deployment of AI systems — a reduction of over 90% — generating a documented financial effect of $2.3 million in savings during the programme period, with a projected total of $3.7 million by year-end. These figures represent a direct return on the investment in AI infrastructure and demonstrate that the efficiency gains achievable through AI deployment at this scale are material rather than marginal. 

The broader context in which these savings are realised is equally important. As TBC Uzbekistan's customer base expands and the range of digital financial products grows, the volume of customer interactions is set to increase substantially. The rising consumer interest in digitally accessible financial services — reflected in the growing demand for products such as "kreditga mashina olish" and "автокредит ташкент" — signals a broader shift toward digital-first financial decision-making that TBC Bank Uzbekistan's AI platform is structurally equipped to support efficiently and at scale. 

Uzbek-Language AI Capabilities Address a Structural Market Gap 

Among the programme's most distinctive technical achievements is the development of a custom large language model trained on the Uzbek language, incorporating both automatic speech recognition and text-to-speech capabilities. The creation of an Uzbek-language LLM addresses a structural gap in the regional AI landscape: the absence of high-quality, locally trained language models capable of handling the nuances of Uzbek-language customer interactions at production quality. This capability is not merely a technical milestone; it is a strategic asset that enables TBC Uzbekistan to serve its customer base in their primary language with the same AI-driven efficiency applied to Russian and English interactions. 

The development of internal data quality benchmarks alongside the language model ensures thfat the systems built on this foundation meet the reliability standards required for use in a regulated financial environment. These benchmarks establish a baseline against which future model improvements can be measured, creating a framework for continuous advancement rather than a one-time technical achievement. 

AI-isation Programme Embeds Adoption Across the Organisation 

Parallel to the technical build-out, TBC Uzbekistan executed an AI-isation programme designed to transform how employees across every business unit interact with AI tools in their daily work. The objective was to ensure that AI adoption extended to commercial, operational and support functions throughout the organisation — not only the technology and data teams responsible for building the platform. This cultural dimension distinguishes TBC Uzbekistan's approach from organisations where AI investment remains confined to isolated use cases without broader institutional adoption. 

The programme's long-term ambition extends beyond TBC Uzbekistan's internal operations. The AI platform has been architected to function not only as an internal capability but as a potential AI service provider for companies within the broader TBC holding — a positioning that transforms the investment from a cost efficiency initiative into a potential revenue-generating platform. As TBC Uzbekistan continues to scale its ecosystem across consumer banking, business services and digital payments, the AI platform developed in 2025 provides the operational foundation on which the next phase of growth will be built.