Incremental AI Formula Detection in BOE

Feature Request: Incremental AI Formula Detection in BOE

Summary:
Enable the AI formula detection capability in the BOE to recognize and process only newly added formulas when rerun, without reprocessing or altering formulas that were previously detected and confirmed. This creates an incremental detection workflow that preserves existing results while capturing changes.

Problem Statement:
Today, when users run the AI formula detection on a BOE and later add more formulas or text, rerunning the tool risks reprocessing previously detected formulas, causing unnecessary review effort and potential confusion. Users need a reliable way to capture only the newly added formulas without disturbing or duplicating the original detection results, especially on complex or frequently updated BOEs.

Proposed Solution:
Update the AI formula detection logic so that, after the initial parse, subsequent runs only detect formulas that were added or modified since the last successful detection.
Track which text/formulas have already been identified and confirmed and treat them as “locked” or non-editable during subsequent parses unless the user explicitly chooses to re-parse everything.
Provide a clear option in the UI to:

• Run “Incremental detection” (default) that only finds new formulas.

• Run “Full re-detection” when the user intentionally wants to rescan the entire BOE.
  Ensure that existing formula mappings, metadata, and connections remain intact during incremental runs.
  Offer feedback (e.g., count and list of newly detected formulas) so users can quickly review only the incremental changes.

Benefits:
Reduces rework by avoiding repeated review of previously detected formulas.
Improves accuracy and stability of BOE data by preserving existing formula mappings.
Saves time for users working on large or evolving BOEs with frequent edits.
Provides a clearer, more predictable AI workflow through incremental detection and explicit control over full re-parsing.
Enhances user confidence in AI-assisted formula detection by minimizing disruption to prior results.