TRUSTBANK uses AI agents to personalize Furusato Nozei gifts

Japanese donation platform TRUSTBANK has launched Choice AI, a conversational recommendation agent built in partnership with Recursive using OpenAI models. The agent personalizes gift discovery for the Furusato Nozei system, where taxpayers can donate to municipalities and receive gifts in return. By engaging users in natural dialogue, Choice AI understands preferences and suggests relevant local products, streamlining what was previously a manual search process. For developers building AI workflows, this case demonstrates how large language models can be integrated into e-commerce flows to handle open-ended user intents, generate personalized options, and reduce decision fatigue. The implementation also highlights the importance of domain-specific fine-tuning and careful prompt engineering to align recommendations with both user desires and the constraints of the donation system. Builders can study the architecture to apply similar pattern-matching and conversational retrieval in their own recommendation engines.