KnowPlan AI Agents Optimize University Degree Pathway Planning.

Planning a university degree is a complex task, requiring students to navigate disparate and often unstructured information from various university sources like catalogs, departmental pages, and PDFs. Current planning tools often couple data extraction with planning, leading to inefficiencies and hidden failure modes. KnowPlan addresses this by separating the process into two distinct, certifiable stages. The first stage, CatalogBrowse, is an AI agent that autonomously explores university sources to reconstruct the curriculum. It uses a novel scoring system for actions and a robust parsing mechanism, terminating only when a "closure certificate" confirms data completeness. This stage outputs three provenance-linked JSON documents. The second stage, DegreeMap, consumes only these structured documents. It compiles them into a typed requirement hypergraph and uses a CP-SAT solver to optimize student-specific paths, considering prerequisites, constraints, load and risk, personalized utility, and option value. This two-stage approach ensures high data recall (96.2%) and significantly improves personalized utility in degree planning, demonstrating a highly effective and certifiable solution.