Post by Chemlali Rayen

Planning a trip today is overwhelming: hundreds of flights, conflicting reviews, visa rules, budget constraints. And once you book everything, you are on your own. No one watches your flight status at midnight. No one tells you the museum is closed on Tuesdays. What if a system could plan your trip intelligently and stay with you through the entire journey? That is what 𝗥𝗮𝗳𝗶𝗾 𝗔𝗜 is. Our end-of-year project, an AI travel planning and companion system built on a multi-agent architecture. Rather than one model doing everything, we built specialized agents each owning a specific domain, coordinating through a shared structure without ever talking to each other directly. The result is a system that is modular, consistent, and easy to extend. The orchestration is where things get interesting. The system understands your intent, validates every decision through a dedicated Critic, and corrects itself when something does not add up. Change your mind mid-plan? It figures out exactly what changed and continues from there, without starting over. Three innovations made the system genuinely smart: 𝗢𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘀𝘁𝗶𝗰 𝗥𝗔𝗚 The system has no pre-built knowledge. It builds its own, in real time, relevant to your exact trip. Always fresh. Never stale. 𝗧𝗵𝗿𝗲𝗲-𝗟𝗲𝘃𝗲𝗹 𝗠𝗲𝗺𝗼𝗿𝘆 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 It remembers you. Not just for this session, but across time. The more you use it, the better it understands your travel style and preferences. 𝗦𝗽𝗲𝗰𝘂𝗹𝗮𝘁𝗶𝗼𝗻 𝗘𝗻𝗴𝗶𝗻𝗲 It thinks ahead. Most of the time, by the time you ask a question, the answer is already ready. 𝗖𝗼𝗺𝗽𝗮𝗻𝗶𝗼𝗻 𝗔𝗴𝗲𝗻𝘁 It listens to what is happening around your trip in real time and proactively alerts you when something changes. The planning does not stop when you board the plane. A demo is worth more than words. Here is Rafiq AI in action: https://lnkd.in/dqE8dNrG Built with my friends Khalil Ghimaji Hiba Chabbouh, and Leith Engazzou, under the supervision of Mrs. Lilia Sfaxi. I want to sincerely thank her for her guidance, her precise feedback, and her support at every stage of this project. She pushed us to go further than we thought we could. And to Mrs. Sana Hamdi, thank you for your thorough review and insightful feedback during the defense.