Prevent misunderstanding with a context-specific AI interpreter

Challenge: Amsterdam: Wildcard – How can an AI solution solve a challenge faced by a marginalised group?
Target Group: Migrants

To support the Municipality of Amsterdam in helping over 8,000 statusholders each year, Mokin.nl is building an AI interpreter: a real-time speech-to-speech translation app designed for public service conversations.

The goal is to reduce misunderstandings between municipal staff and statusholders, especially when no translator is available. Unlike generic tools, the app uses context-specific words like “bijstandsuitkering” or “belastingdienst,” which are hard to translate and often unknown to newcomers.

Co-creation is a key part of the project. Together with statusholders, the team collects important Dutch words and concepts to train the AI. A roleplaying workshop was held with a statusholder and municipal staff to test real-life situations. Feedback from this session helped improve the app’s language use and features.

The pilot shows better understanding in conversations and more confidence from both sides. The updated word list, new language options, and privacy-aware features all contribute to smoother communication.

Next steps include further testing in Amsterdam and preparing to scale the tool across Europe, where it can support integration for over one million refugees every year.