TRAJEXA: Evidence-Bounded Personal Exploration from Past Experience to Untried Possibilities

People repeatedly have to choose among possibilities they have not yet experienced: what to study at university, whether to continue formal education or enter work, which first job to pursue, whether to change career, what hobby to try, or what kind of training to undertake. Direct evidence for these choices is often unavailable, yet converting interests, personality, academic history, or prior job titles directly into “fit” can mistake lack of exposure for lack of ability and can restrict future possibilities to labels already present in the past. This preprint introduces TRAJEXA, an evidence-bounded personal exploration system that preserves distinctions among self-ratings, interests, short task responses, experiences recalled through multiple cues, participant-labelled operations, imagined counterfactuals, and future trials. Rather than compressing a person into one score or one representative life episode, TRAJEXA identifies operations that reappear across separate experiences and preserves provenance from derived hypotheses back to their supporting evidence. The system uses an Experience Structure Graph with typed relations and explicit source kinds. Visual distance and clustering are treated as interface layout rather than measures of ability or psychological similarity. Imagined counterfactuals are kept separate from lived experience, and future trials are recorded separately from retrospective recollection. The paper proposes a broader application framing: university-major and education-pathway exploration, first employment and career change, hobby exploration, and workplace training can be treated as instances of a shared problem - using operations evidenced in past experience to explore environments not yet tried. Application examples are candidate contexts for small trials, not predictions of fit or success. The current contribution is a design and research-software result supported by synthetic counterexample auditing. It does not establish psychometric validity, occupational or educational prediction accuracy, stable latent abilities, or the optimality of any recommended path. Its demonstrated claim is narrower: heterogeneous personal evidence can be kept semantically distinct while still generating concrete, traceable possibilities for what a participant might try next.