How the fit engine works
What it reads, how it reasons — and what it cannot tell you
What it reads
What you tell Ora
Your side of the conversation, and the stories when you play them
Assessments you upload
Reports you already have — Big Five, RIASEC, CliftonStrengths and similar
What matters to you
Where you want to be, who you want around you, what your family can afford
How the pieces fit together
Each source is normalised into the same set of named attributes, so any reading can say which inputs moved it. Some things are excluded by design and never inferred at all: your religion, your health, and your politics.
How it reasons
There is no machine learning in here, and we would rather say so than dress up arithmetic. Each side of fit is worked out from named inputs using weights we can show you and change — not weights learned from anyone's data. The code calls its own personality component a heuristic, and that is the right word for it.
What we have not validated
Nothing, yet — and that is the honest answer. We have not followed a student from a reading through to how things actually turned out, so we have no accuracy figure and we are not going to quote one. There is no set of known-good answers behind the engine today. Until there is, treat what it tells you as a structured second opinion, not a measurement.
Explainability
Two things are true at once, so here are both. The reasoning does exist: readings record which inputs moved them, and every Fit Check answer is built to carry its reason in words — why a college might suit you, or why not. But the service that runs the live calculation still hands back numbers with that reasoning detached; joining the two is being built now. We would rather tell you which half is finished than imply every answer already arrives explained.