DocentPro is developing an AI travel platform that aims to streamline the entire travel planning process, from discovery to booking, by creating a modular multi-agent system using LangGraph and LangSmith. This approach allows for the creation of reusable agents that work across different trip planning and real-time conversation domains, such as attractions, restaurants, hotels, and activities. To balance flexibility with control, DocentPro uses a hybrid approach combining LLMs with deterministic logic to ensure practical plans that users can follow. The system is also designed to be observable and debuggable using LangSmith, which provides visibility into the decision-making process. Additionally, DocentPro has successfully integrated support for audio guides in 12 languages using LangGraph, enabling the creation of structured, per-language audio playlists. The platform's current state combines modular agents with deterministic algorithms, traceability via LangSmith, and powers AI itineraries and multilingual audio guides for travelers worldwide.