Customer story · FintechThe MetropolCard Assistant that makes card services faster.
MetropolCard, Turkey's first environmentally friendly digital meal card, serves over 400,000 daily active users and more than 50,000 affiliated businesses. Together with CBOT, it built the MetropolCard Assistant on OpenAI's GPT-4o, backed by a RAG infrastructure over MetropolCard's own documents, to answer user questions accurately and hand off to live support when needed.
MetropolCard, Turkey's first eco-friendly digital meal card, worked with CBOT to absorb the repetitive demands of a fast-growing user base instead of piling them onto the call center. Built on GPT-4o and fed by a RAG layer over MetropolCard's own documents, the assistant answers accurately and consistently. This lowers the carbon footprint, frees agents for higher-value work, and sets a scalable standard aiming at 1 million conversations a year.
With over 400,000 daily active users and more than 50,000 affiliated businesses, MetropolCard needed to meet user demand at the pace of its own growth. Routing repetitive questions (card usage, balance inquiries, membership processes) to the call center lengthened wait times, deprioritized more critical support needs, and increased costs.
Over 50,000 affiliated businesses across 81 provinces and 922 districts, more than 1 million physical card users and over 400,000 daily active users meant demand had to be met at the same pace.
Repetitive questions (card usage, balance inquiries, membership processes) were being routed to the call center.
Using agent resources for simple requests pushed more critical support needs into the background and lengthened wait times.
As Turkey's first environmentally friendly digital meal card, the goal was to move service to a sustainable, technology-driven standard.
MetropolCard partnered with CBOT to build the MetropolCard Assistant on OpenAI's GPT-4o. MetropolCard's own documents, content and information sources were processed with custom embedding models and integrated into a Retrieval-Augmented Generation (RAG) infrastructure, so the assistant answers accurately using both natural language understanding and MetropolCard's institution-specific knowledge, and directs users to Live Chat whenever needed.
The MetropolCard Assistant was developed on OpenAI's latest model, GPT-4o.
MetropolCard's documents, content and information sources were processed with custom embedding models and integrated into a RAG infrastructure, so the assistant answers accurately using both NLP power and institution-specific knowledge.
It directs users to the Live Chat channel whenever needed, creating an end-to-end customer experience.
Built on GPT-4o and a RAG infrastructure, the MetropolCard Assistant answers topics like card usage, balance and membership on mobile, and hands off to live support when needed.
By moving traditional call center processes onto a digital platform, MetropolCard reduced its carbon footprint, raised customer satisfaction, and freed agents to focus on more complex, valuable issues.
The collaboration set a sustainable, scalable, technology-driven standard for user experience, with a goal of 1 million conversations per year.
More case studies
Every organization's needs are different, but the impact of a well-designed AI experience is shared: faster processes, easier service and stronger customer engagement. Explore the other success stories we have brought to life across different industries.
Osmanlı Yatırım | DirektSor
Integral Yatırım | İntegral Asistan
