EDUPATH AI A Rag-Based Multi-Agentic AI Framework for Personalised Education and Career Advisory

Choosing a career is a tough decision to make in itself. Things get trickier when it comes to information a student actually needs: which scholarships they're eligible for, where courses lead, where students can get in based on their marks, etc., and all that information is spread out across dozens of government portals, college websites, and PDFs that don't always agree on what they're offering. The majority of the career-guidance tools that are available today are not really a solution to this. They either give a general response from a brief questionnaire, or release a chatbot which will freely associate a response that seems confident but is not true. Neither verifies whether a student is qualified or not for what it suggests. This paper is an attempt to bridge that gap by employing a Retrieval-Augmented Generation (RAG) pipeline to achieve this using a set of cooperating agents, instead of a single large model performing all the tasks. The first step is to process a student's academic record, interests, financial status, location, class and career category, and goals to create a structured profile, which is then fed into agents that search for verified information, validate eligibility for scholarships based on hard rules, not assumptions, and compile a plan that is concrete and easily explained. We evaluated the retrieval layer on ten representative queries, including career paths, welfare schemes, higher education and scholarships, and it achieved a top five correct source hit rate of 90% and a Mean Reciprocal Rank of 0.7167. To sum up, if you base the model's output on retrieved evidence and ensure that eligibility decisions are deterministic then you end up giving guidance that a student can actually trust and follow — something that a pure language model and a static rule engine can't do alone.