Our goal was to use a virtual assistant to give fast recommendations based on customer requests or interests. We combined IBM's Watson Assistant with Neo4j graph database to achieve this. The system works by asking questions in an order determined by edge weights in the graph, updating the graph based on user responses, and making recommendations. This approach allows for interactive and bidirectional dialog, enabling the virtual assistant to provide useful guidance and helping users find the best answer quickly and efficiently. We believe this technology can be applied to various fields such as sales, medical diagnosis, technical support, and more, with graphs tailored to specific topics or themes. By making virtual assistants reactive and adaptive, we can enhance their capabilities and provide valuable advice to users.