Large Language Models (LLMs) have transformed automated trip planning by enabling the proliferation of these models. Nevertheless, current systems are mostly text-oriented, which restricts users, who require visual media as their source of inspiration. This paper presents WanderSnap, a full-stack, Artificial Intelligence AI-based travel planning app, which combines multi-modal input features, so that users can make queries through text or image prompts. We introduce the system architecture that uses the Gemini Application Programming Interface API of Google, React, and Firebase. We performed within subjects user study (N = 32) to compare our multi-modal system to a text only baseline. We find that the multi-modal approach has a statistically significant increase in user satisfaction ( \(p<0.01\) ) and perceived itinerary quality ( \(p<0.05\) ), as well as a 28% per task completion time. These discoveries imply that visual input is an important consideration that must be considered to develop more intuitive and useful AI travel assistants.