How Yettel Hungary Is Using AI to Transform Customer Care - Without Losing Control of Its Data

The telecommunications industry is defined by a paradox of scale: the more customers you serve, the harder it becomes to manage the ocean of documentation required to support them. On any given day, thousands of customers reach out via phone, email, or digital chat seeking answers to complex technical questions or simple billing queries. For companies operating in this space, the primary challenge is not a lack of information, but the speed at which that information can be accessed and applied by large support teams. When knowledge is scattered across fragmented systems, it creates a fundamental friction that impacts both the employee experience and customer satisfaction.

Solving the knowledge fragmentation problem

The journey into artificial intelligence for a major telecom provider is often driven by a need to organize this internal complexity. In our recent conversation with Denitsa Gavrilova, Director of AI and Data at Yettel Hungary, she explained that their journey was a strategic necessity. As Denitsa noted: "It all started a couple of years ago when we decided to have a huge strategic focus on digitalization. Now with AI, it is a natural evolution of the digital part".

Yettel Hungary, which has evolved from its 1994 roots as Pannon GSM to its current standing in the e& PPF Telekom Group, identified a clear business case for AI within their customer care department. The primary hurdle was a fragmented knowledge base that made it difficult for agents to onboard quickly or provide rapid resolutions for generic queries. By automating routine functions, Yettel aimed to allow human agents to focus on the most complex customer needs.

The choice between in-house development and a strategic partner

When deciding how to build this infrastructure, many enterprises face a choice between building entirely in-house or buying a generic solution. Denitsa Gavrilova recognized that in-house development often creates a bottleneck because the pace of AI innovation is faster than traditional enterprise cycles can manage. Conversely, relying solely on an off-the-shelf product makes a company too dependent on a vendor’s specific roadmap, limiting their ability to navigate toward unique strategic goals.

The solution was a blended approach: finding a reliable AI brain to act as a modular engine within their custom architecture. They selected Iris.ai to provide this technical layer, specifically utilizing agentic RAG (Retrieval-Augmented Generation) modules. This allowed them to deploy a working solution in weeks that prioritized technical capability and substance over simple generative fluency.

What happens when an AI product goes live

The transition from development to a live environment revealed that AI product management follows a very different lifecycle than traditional software. In standard software projects, the work often winds down after deployment. However, Denitsa emphasized a critical shift in mindset: "The work actually starts when the product is launched".

This is because natural language is inherently unpredictable, and internal test cases rarely mirror how real customers speak. Working closely with the expertise at Iris.ai, the team learned that the true development of an AI product is driven by the feedback and reality checks that occur in production. This collaborative process transformed Yettel's internal understanding of how to manage an AI product, moving away from rigid planning toward a more fluid, feedback-driven evolution. You can see more examples of this transition in our Case Study.

Building a compounding strategic asset

Looking toward the future, the partnership model is designed to create compounding value. In the world of customer care, switching vendors for every new project means starting over with legal, security, and procurement procedures, which slows down the entire organization. By maintaining a strategic partnership, every new use case can build on the existing data and technical knowledge already established.

This approach allows Yettel Hungary to maintain full control and knowledge of its data, which is essential for managing sensitive business information. As Denitsa Gavrilova summarized, having a strategic partner like Iris.ai means that "with the next use case we are not starting from scratch". It ensures that the company remains at the cutting edge of digital transformation while adhering to strict standards for information security and business continuity.

The road ahead for deep knowledge

As the industry moves toward more autonomous, agentic systems, the organizations that will lead are those that have already unified their knowledge foundations. The collaboration between Yettel Hungary and Iris.ai is a testament to what happens when technical precision meets a long-term strategic vision. By focusing on factuality and smart algorithms, Yettel is setting a new standard for how telecommunications companies can serve their customers in the digital age.

You can also listen to our conversation with Denitsa Gavrilova from our last LinkedIn Live here: https://www.linkedin.com/events/7457056650605195264