AI Chatbot for a Japanese Used Car Exporter
An English AI customer service agent built on RAG + LLM and embedded in the client website. It handles about 3,000 overseas inquiries a month and cuts reply time per inquiry to about one third.
Challenge
A Japanese used car exporter serves customers mainly in African countries, and inquiries pour in through its website in English. For every inquiry, customer service staff had to check inventory, look up vehicle specifications, and verify vehicle condition before writing a reply in English. Reply quality varied from person to person, searching vehicle information took time and effort, and inventory checks were entirely manual. As inquiry volume grew, the team's response speed and consistency hit their limits.
Solution
Scope of work: Design, development, and launch of an AI chatbot built on the company's internal data.
- Knowledge sources: Internal inventory data, onboard vehicle data, vehicle specification sheets, and historical operations records connected to the retrieval layer, so answers are grounded in the client's own data rather than the generalities of a generic model.
- Architecture: A retrieval-augmented generation (RAG) architecture first retrieves vehicle and inventory information relevant to the inquiry from internal data, then a large language model (LLM) writes the reply, balancing accuracy with natural wording.
- Language and channel: Replies in English for overseas buyers in African markets. The chatbot is embedded directly in the client website, so buyers can start a conversation on the same page where they browse vehicles.
- Delivery timeline: About six months of development, launched in 2025.
Results
- About 3,000 inquiries a month: Since launch, the chatbot handles most of the English inquiries coming through the website.
- Reply time per inquiry cut to about one third: The research and English writing time needed for each inquiry dropped to about 1/3 of what it was before, and reply consistency improved as well.
- Reusable for export trade: "Internal data + RAG + multilingual replies" is the same answer to the same problem for any exporter whose inquiries come mostly in foreign languages.
What the client said
"The AI chatbot helps us reply to customers faster and more consistently, and our staff spend much less time searching for vehicle and inventory information."
See other case studies
Eight platform-wide sales events in over two years with zero major incidents, load tested at up to 8 times normal traffic.
Video data and transaction data connected across about 500 stores, cutting the investigation of a suspicious transaction from a week to a day.
