How SLTMobitel transformed urban tourism with its GSMA-facilitated AI-powered Virtual City Assistant

Executive summary

As global tourism evolved towards more digital and personalised experiences, and the concept of the ‘smart city’ gained momentum, SLTMobitel identified a critical opportunity to move beyond its role as a core connectivity provider. The company sought to become an enabler of Sri Lanka’s digital transformation by creating value-added services that could enhance the urban experience for both tourists and citizens. The urgent need to offer a modern, seamless, and accessible information ecosystem for visitors prompted SLTMobitel to explore the capabilities of generative AI. This strategic move was designed to meet the expectations of modern travellers and position Sri Lanka’s cities as technologically advanced and visitor-friendly destinations.

The Challenge

Before the implementation of the Virtual City Assistant, the process of delivering information to tourists and citizens was inefficient and faced several key hurdles. These challenges limited the ability of tourism authorities and municipal bodies to engage effectively with their target audiences.

• Business Challenges:
  • Fragmented Information: Critical information for tourists was scattered across physical brochures, multiple websites, and time-limited information desks, leading to an inconsistent user experience.
• • Language Barriers: Engaging with a diverse, international tourist base was difficult, as providing real-time support in multiple languages was not operationally feasible.
• • High Operational Costs: Relying on physical infrastructure and staff for visitor support was costly and did not scale effectively during peak tourism seasons.
  • Lack of 24/7 Availability: Tourists arriving at unconventional hours or needing information outside of standard business hours had no reliable source of support.
• Technical Challenges:
  • Data Silos: Important civic and cultural information was locked away in unstructured documents (PDFs, Word files), Portfolio websites and databases, making it difficult to access and query in real time.
  • Lack of Service Integration: There was no unified platform to integrate disparate but essential services, such as maps for navigation, booking platforms for accommodation, and cultural FAQs.
  • Content Management Inefficiency: Updating information across all channels was a slow, manual process, leading to the risk of outdated information being shared with the public.

“We recognised that to truly create a smart city, we needed to bridge the information gap for our visitors. The challenge was not just providing data, but delivering it in an interactive, accessible, and multilingual way, 24/7.”

Anil Pradeep Kumara, Head of AI & Data, SLTMobitel

The solution: a collaborative approach

• The Process: The collaboration was catalysed through the GSMA Foundry, where SLTMobitel became one of the first mobile network operators to begin their AI evaluation using the GSMA and IBM GenAI Accelerator Pack. This initiative provided the framework and initial tools for SLTMobitel to start building proof-of-concept (PoC) trials using generative AI. Building on this foundation, the The Embryo Innovation Team at SLTMobitel defined a clear PoC: the development of a Virtual City Assistant that would serve as a single, intelligent point of contact for tourists.
• The Technology: The solution was built using an enterprise-grade AI and data platform, designed to build, tune, and deploy sophisticated AI models. Its core strength lay in its ability to build multilingual conversational AI agents. The platform’s Retrieval-Augmented Generation (RAG) capability was central to the FAQ Agent; it allowed the assistant to provide accurate, context-aware answers by drawing directly from a verified knowledge base of documents and URLs. This approach ensured that the assistant’s responses were grounded in factual, approved information. Furthermore, the platform’s architecture enabled seamless integration with external APIs, which was critical for powering the Route Agent (using Google Maps API) and the Booking Agent (linking to third-party accommodation platforms).

SLTMobitel’s expertise

As the national ICT solutions provider, SLTMobitel was uniquely positioned to drive this transformative project.  The successful implementation of the Virtual City AI Chat Assistant hinged on several of SLTMobitel’s core competencies, demonstrating their technical prowess and commitment to enhancing Sri Lanka’s digital landscape.

Key to the project’s success were the meticulous UI adjustments made to ensure a seamless and intuitive user experience for tourists and citizens alike. Leveraging their deep understanding of user-centric design, SLTMobitel’s team crafted an interface that was not only easy to navigate but also catered to a multilingual audience, breaking down communication barriers.

The integration of their Large Language Model (LLM) formed the intelligent core of the Virtual City Assistant. This sophisticated AI was trained to understand and process natural language queries with a high degree of accuracy, enabling it to provide relevant and contextual information. SLTMobitel’s expertise in AI and data platforms was crucial in deploying a solution that could handle a diverse range of requests, from simple FAQs to more complex booking and routing inquiries.

Furthermore, the development of a comprehensive knowledge base built according to their specific requirements was a critical component of the solution. SLTMobitel worked closely with tourism stakeholders to aggregate and structure vast amounts of information, ensuring the Virtual City Assistant was a reliable and up-to-date resource. This bespoke knowledge base, combined with the power of the LLM, allowed the assistant to deliver personalized and actionable insights to its users. Through these concerted efforts, SLTMobitel has not only showcased its technical capabilities but has also reinforced its role as a key enabler of Sri Lanka’s digital transformation.

Implementation and deployment

The project was rolled out as a phased PoC to validate the concept and technology before moving toward a full-scale deployment at the end of 2025. The implementation began with the creation of a secure knowledge base, where documents were chunked, embedded, and indexed into a vector database. The development team then built and trained the three distinct agents (FAQ, Route, Booking), ensuring each could perform its specialised task reliably. A key phase involved the integration with third-party APIs for maps and booking, with a strong focus on secure data handling. Finally, a mobile-first user interface was developed and tested to ensure a responsive and intuitive experience for end-users on any device. This agile methodology allowed the team to learn and adapt throughout the deployment, ensuring the final PoC was robust and fit for purpose.

Results & business impact

The pilot of the Virtual City Assistant successfully demonstrated its transformative potential for enhancing urban tourism and civic engagement.

• Quantitative Impact (The Numbers):
  • The proof-of-concept established the technical viability and laid the groundwork for future measurements, such as reduction in inquiries to physical tourist centres and increased engagement with digital content.
• Qualitative Impact (The Strategic Benefits):
  • Improved Customer Experience: The assistant offered a seamless, personalised, and instantly accessible journey for tourists, providing 24/7 support in multiple languages and directly addressing a major pain point.
  • Enhanced Agility: The included Admin Portal empowered tourism authorities and municipal bodies to manage and update information in real-time, ensuring that content remained current and relevant.
  • Future-Proofing: This investment created a scalable and secure foundation for SLTMobitel’s digital transformation strategy. The platform was designed to be extensible, allowing for the future addition of new agents and services e.g., public transport updates, event ticketing etc. as the smart city ecosystem evolves.

“This project is a prime example of how the GSMA Foundry can accelerate innovation. By connecting forward-thinking operators like SLTMobitel with cutting-edge technology, we help unlock the potential of AI to solve real-world challenges and create immense value for society.”

Richard Cockle, Head of the Innovation Foundry, GSMA

The future outlook

Following the successful pilot, SLTMobitel plans to move towards a full rollout of the Virtual City Assistant, initially targeting a major urban area before expanding to other cities and tourist destinations across Sri Lanka. The future roadmap includes enriching the assistant’s capabilities by adding more specialised agents and deepening the integration with local businesses to offer more personalised recommendations and services. SLTMobitel will continue its collaboration with the GSMA and other technology partners to explore new applications of AI, reinforcing its commitment to leveraging cutting-edge technology to drive national progress and innovation.