SafarGlimpse shifts from generic AI prompts to a hybrid Graph + RAG engine. Most travel AI tools generate itineraries that ignore real travel flow; we use distance-aware clustering and real-world constraints so plans are actually walkable. By grounding LLMs in structured place data and graph relationships, we reduce hallucinations and produce itineraries that feel local, practical, and thoughtfully designed.

I would love to hear your feedbacks and suggestions