TBC Bank Uzbekistan has concluded a major artificial intelligence transformation programme, completing the build-out of an end-to-end AI platform across its operations between January and November 2025. The initiative, spanning 45,000 man-hours of development effort, delivered measurable reductions in operating costs, a significant expansion of AI-powered customer interactions, and the establishment of a dedicated internal machine learning competence centre — placing TBC Uzbekistan among the most advanced AI-driven financial institutions in Central Asia.
The programme was executed under the leadership of Andrey Malygin, Head of AI Platform at TBC Uzbekistan, and covered technology infrastructure, product deployment, cultural adoption, and data governance simultaneously. Its completion marks a decisive transition from AI experimentation to systematic, enterprise-wide integration that is already generating measurable commercial returns.
From $0.35 to $0.03: The Economics of Automated Customer Service
The most immediately quantifiable outcome of the programme is a reduction in the cost of customer interactions from $0.35 to $0.03 per call — a decline of more than 90 percent. Across the volume of interactions TBC Uzbekistan processes, this shift represents a structural change in the unit economics of customer service rather than an incremental efficiency gain.
By the programme’s close, AI systems had processed over 1.6 million calls and 690,000 chat interactions. The financial effect attributable to these deployments reached $2.3 million in documented savings during 2025, with a projected total of $3.7 million as the full-year impact of systems deployed in the latter part of the year flows through the organisation’s cost base.
These figures establish a compelling commercial case for the investment. At 45,000 man-hours of development effort, the programme’s return profile compares favourably with most technology initiatives in the regional banking sector — particularly given that the infrastructure built is reusable and scalable, designed to support an expanding portfolio of AI products rather than a single narrow application.
Three AI Products Enter Production Across the Customer Lifecycle
Three AI-powered products entered production as direct outputs of the programme. The first is an AI Assistant capable of handling customer interactions across both voice and chat channels — addressing the highest-volume touchpoint in the bank’s customer service operation with consistent quality and without the staffing constraints of human-only delivery.
The second and third products are a Sales Assistant and a Collections Assistant, each targeting a distinct phase of the customer lifecycle. The Sales Assistant is designed to support and partly automate the customer acquisition process, enabling higher throughput without proportional increases in headcount. The Collections Assistant applies AI to the management of overdue accounts — a function where response speed, consistency, and tone calibration directly affect recovery rates and customer outcomes.
Deploying all three products in parallel reflects a deliberate strategy to demonstrate AI’s commercial utility across the full customer journey — from acquisition through servicing to remediation — rather than concentrating initial investment in a single use case and gradually extending scope.
Digital Credit Demand and the Consumer Context Driving AI Investment
TBC Uzbekistan’s AI transformation is unfolding against a backdrop of rapidly shifting consumer expectations in the Uzbek financial market. Sustained growth in search activity around terms such as maqsadsiz kredit olish points to a population increasingly oriented toward digital-first financial access — one that expects fast, low-friction credit products without the documentation requirements and branch visits associated with traditional banking models.
For TBC Bank Uzbekistan, whose AI platform directly accelerates the speed and personalisation of customer interactions, this demand environment represents both a validation of strategic direction and a continuous source of competitive pressure. The bank’s capacity to deploy AI at the point of customer contact — across lending decisions, payments, and service interactions — is central to its ability to meet these expectations reliably and at scale.
Building Uzbekistan’s First In-House Large Language Model
Among the programme’s most technically significant outputs is the development of Uzbekistan’s first in-house large language model, encompassing both automatic speech recognition and text-to-speech capabilities in the Uzbek language. Building language AI for Uzbek cannot be achieved by adapting existing English or Russian-language models — it requires purpose-built training datasets, specialist linguistic expertise, and sustained engineering investment that most organisations in the market have not been willing to commit.
TBC Uzbekistan’s decision to develop this capability in-house, rather than relying on third-party providers, reflects the programme’s central objective: building an AI foundation that is independent of external vendors and fully under the bank’s strategic and operational control. This independence reduces long-term cost exposure, protects data sovereignty, and enables the bank to iterate on its AI capabilities without being constrained by the roadmaps or pricing decisions of external suppliers.
The technical stack underpinning the platform spans PyTorch, TensorFlow, and LangChain for machine learning development; Airflow, Spark, and Kafka for data infrastructure; and MLflow with Kubernetes for MLOps. Monitoring and governance pipelines built on Prometheus and Grafana ensure operational reliability and regulatory compliance across all production systems.
Cultural Transformation and the Path Toward an AI-First Organisation
Alongside the technology build-out, TBC Uzbekistan ran a parallel AI-isation programme designed to embed AI usage into the daily workflows of employees across all business units. The underlying principle is clear: AI infrastructure delivers its full value only when the organisation using it has the capability and motivation to leverage it effectively. Technical deployment without cultural adoption produces underutilised systems and returns that fall well short of potential.
The establishment of an in-house AI team and a dedicated ML Competence Center is the institutional expression of this commitment. Rather than treating AI as a narrow specialist function managed by a small central team, TBC Uzbekistan has invested in building distributed AI capability — enabling individual business units to identify opportunities and contribute to product development without complete dependency on a central engineering group.
The platform has also been designed with the capacity to serve other entities within the wider TBC holding, functioning as an AI service provider across the group. This architecture extends the return on the infrastructure investment beyond TBC Uzbekistan’s own operations and creates a platform for ongoing innovation that will shape the bank’s competitive position for years to come.
