login_FILL0_wght400_GRAD0_opsz24 MYEURESA
browse all use cases
use cases

New Mathias: Customer‑Centric Conversational AI for Enhanced Digital Journeys

  • by Ethias (Belgium)

Objectives

New Mathias aims to deliver a modern, reliable and customer‑centric conversational assistant that enhances the digital experience while reinforcing Ethias’ leadership in responsible AI.

  • Deliver clearer, more accurate and more personalized guidance to customers and prospects.
  • Improve engagement and simplify digital journeys by answering insurance questions instantly and safely – 24/7.
  • Reduce operational workload by handling recurring, low‑value requests.
  • Build a scalable AI foundation to support future applications such as the Claims Assistant.
  • Guarantee controlled, transparent and trustworthy responses in line with the EU AI Act.

Progress status

Completed / Deployed
  • The 1st public version of New Mathias is live since December 2025.
  • Throughout 2026, new capabilities will progressively enrich the assistant while ensuring full alignment with responsible AI standards.

Work team

  • Developed in house
  • Project Management: Project mgmt.: Led by AI, Innovation & Automation Dept. (Digital Transformation Office) with Customer Experience & Business teams.
  • External partner/team: Developed with NRB (Ethias IT subsidiary)

Work carried out to date

The first GenAI‑powered version was successfully released in December 2025. Since then, Ethias and NRB have:

  • Consolidated the 2025–2026 roadmap,
  • Deployed key enablers (evaluation framework, monitoring, cloud foundations),
  • Improved performance and UX to support upcoming extensions.

This provides a robust and future‑proof base for the next stages of value creation.

First Results & Lessons learnt

  • Early observations show encouraging results, with increased conversation volumes and reduced negative feedback. Customers already benefit from more accurate and helpful answers.
  • The new GenAI architecture also demonstrates high agility: new products and updates have been integrated with minimal additional effort.
  • Enhanced reporting now provides better insights into customer intents, helping prioritize future improvements
  • Overall, New Mathias confirms that a modern and responsible AI foundation reduces maintenance efforts while enabling more impactful digital experiences.

Difficulties encountered & Remainng Challenges

Key challenges still ahead include:

  • Testing GenAI at scale to ensure consistent and safe responses for a wide variety of customer questions.
  • Maintaining alignment with the EU AI Act, especially as new capabilities are added.
  • Preparing secure extensions, such as future integration into authenticated environments.
  • Adapting to evolving LLMs, which requires strong automated validation and regression testing capabilities.

Interested? Hear more about this use case during Workshop 3