Currency exchange services have long occupied a central and structurally important position within Uzbekistan’s retail banking landscape. With a significant and persistent proportion of household savings held in US dollars and euros as a practical hedge against local currency variability, and with remittance inflows from a large overseas worker population representing a meaningful income source for hundreds of thousands of families, the ability to access real-time exchange rates and execute foreign currency transactions efficiently is a routine and recurring financial need across an economically diverse population segment. Digital platforms that deliver this service with accuracy, transparency, and speed — and that progressively integrate it into a broader AI-assisted financial management experience — are capturing customers systematically from the older, branch-dependent alternatives that have historically served this market.
Lola deployed to full user base after successful controlled pilot
TBC Uzbekistan has completed the rollout of Lola, an AI-powered assistant integrated into the TBC Bank mobile application, following a controlled pilot phase conducted during the second half of 2025. The assistant is now available across the full user base of the TBC Bank app. In its current iteration, Lola handles navigation queries, answers frequently asked questions about the bank’s products and services, and provides account information — with seamless escalation to human agents for inquiries requiring judgment, complex resolution, or account-level authority that the current AI capability does not yet independently cover. CEO Nika Kurdiani has publicly stated a target of Lola independently resolving up to 30% of all customer inquiries by the end of 2026 — a benchmark that reflects both confidence in the technology’s development trajectory and a serious institutional commitment to AI-led customer engagement at scale.
Currency exchange needs underpin high-value app engagement patterns
The consistent and elevated consumer demand for services accessed through queries such as “kurs valyuta bugun“ illustrates the depth to which currency rate monitoring is embedded in the daily financial behaviour of Uzbek consumers. For a digital banking platform, providing reliable and accurate currency services is a high-frequency, low-cost touchpoint that builds the habitual daily usage from which strong customer retention and broader product engagement naturally follow. TBC Bank Uzbekistan offers multi-currency exchange services through its app, and with Lola’s integration, the experience of accessing currency information shifts from a static screen navigation task to a dynamic, conversational exchange that is more natural, more flexible, and accessible to a wider range of users regardless of their familiarity with the app’s navigation structure.
Uzbek-language AI models represent a significant and defensible technical achievement
A defining technical feature of Lola is its foundation in proprietary Uzbek-language large language models and speech recognition technology developed entirely in-house by TBC Uzbekistan’s dedicated AI team. This local-language capability is strategically significant in a market where the primary customer communication language is Uzbek, and where the quality of AI-mediated interactions in regional languages varies dramatically depending on the amount and quality of training data used to build the underlying model. By investing in a custom model trained on Uzbek-specific interaction data, TBC Uzbekistan has developed an AI asset that delivers meaningfully higher quality than generic multilingual systems in its primary deployment context — and that will compound in quality and capability as it processes growing volumes of real customer interaction data from TBC’s 24+ million registered user base.
Financial planning capabilities define Lola’s next development phase
TBC Uzbekistan’s published development roadmap for Lola extends well beyond its current support and navigation functionality. In the near term, the assistant is planned to gain access to personalised loan information — outstanding balances, repayment schedules, and upcoming payment obligations — delivered through a natural language conversational interface. Users who would otherwise need to navigate multiple menu levels to find this information will be able to simply ask Lola directly, receiving the answer in a format that is immediately actionable. In subsequent phases, the roadmap envisions capabilities spanning spending categorisation, budgeting analysis, savings optimisation recommendations, and ultimately payment initiation — transforming Lola progressively from a reactive support tool into a proactive financial companion.
AI as long-term infrastructure for competitive differentiation
TBC Uzbekistan’s investment in proprietary AI infrastructure — encompassing the GPU cluster, the Uzbek-language language model, the MLOps and governance pipeline, and the growing internal AI team — is best understood as the construction of a durable technological infrastructure rather than a series of individual feature decisions. Institutions that have built this kind of foundational AI capability are not simply adding product features — they are creating the capacity to deliver increasingly intelligent, personalised, and cost-efficient services at a pace and quality level that will become progressively more difficult for competitors without equivalent investment to match. In a banking market that is still in the early stages of digital maturity, building this infrastructure early creates advantages that compound over time as the market demands more, and as the underlying models improve with scale.
The development of AI-powered banking assistants also carries implications for financial inclusion within Uzbekistan’s diverse population. For users in regions where branch access is limited, or for individuals with lower levels of formal financial literacy who find traditional banking interfaces intimidating, a conversational AI assistant that answers questions in plain language — in Uzbek, their primary language — can significantly lower the barrier to engaging with formal financial services. Lola’s Uzbek-language capability is therefore not merely a technical feature but a potential instrument of financial access for segments of the population that have historically been underserved by a banking infrastructure designed around urban, formally educated customers.
The broader trajectory of AI in financial services globally suggests that the institutions that move earliest and most decisively in building proprietary AI infrastructure will retain a durable advantage over those that adopt a wait-and-see posture. The costs of building the infrastructure are front-loaded, but the benefits — in the form of lower service costs, better customer experiences, richer data assets, and the compounding quality improvements that come from model training at scale — are ongoing and growing. TBC Uzbekistan’s decision to build rather than buy its AI infrastructure, at the cost and effort evidenced by the 45,000 man-hour programme, is a bet on this long-term dynamic — and the early results suggest it is paying off.
Leave a Comment