AI screen identification & merge
Run LLM screen naming/description over multi-snapshot dynamism only after privacy sanitization masks per-session data (letters→X, digits→9) (raw buffers stay local) then merge duplicate screens by catalog name, identifiers, or structural fingerprint with event repointing.
Raw discovery produces many position-hash screens that are hard to govern. OL TermIQ AI screen identification uses an LLM over multiple snapshots of the same screen (dynamism across values) to propose business-friendly names, descriptions, and field-oriented context, stored as screen identification metadata.