Raw data → metadata → entities → relationships → knowledge graph → understanding.
The ladder from raw data up to understanding: each layer adds structure — metadata, entities, relationships — until the pile becomes a knowledge graph you can reason over.
Each rung buys one specific thing. Raw data is just bytes. Add metadata and it becomes findable. Add entities and it becomes about something — people, places, trips, not just files. Add relationships and it becomes a graph you can traverse. Only then can anything reason over it.
The value of naming the ladder is diagnostic. When AI gives a vague answer about your own material, the useful question isn't "better prompt?" — it's which rung am I missing? Usually the honest answer is entities: a pile of well-tagged files that never became things.
The honest limit: you don't need every rung for every question. Climb until it answers, then stop — over-structuring is its own kind of clutter.