Each demonstration runs entirely in the browser on authored synthetic examples. No model is trained or consulted, no data is uploaded, and every displayed total is recomputed from the fixture in front of you. Every demonstration has a keyboard-operable table equivalent to any graphic.
Temporal comparison workbench
Hold the current study fixed, change the earlier examination or the retrieved reference, and see each stage decide.
Inputs: Selected prior study; retrieved reference set
Outputs: Gate decisions, admission table, realised report, evidence identifiers, stage trace
OpenGraph retrieval and provenance explorer
Ask a question of a small fictional corpus and inspect the typed path from every cited passage to its source.
Inputs: Question; retrieval mode; node-type filters
Outputs: Ranked passages with paths, source drawer, graph view and equivalent table
OpenTuple evaluation and omission explorer
Move an admission threshold over authored candidate tuples and watch precision, recall and report states recompute.
Inputs: Threshold; evaluated fields
Outputs: Tuple metrics, report-state counts, per-report diff, downloadable fixture
Open