VERITAS
Demo workspacenorthgate-consulting
Benchmark

Five numbers, with intervals and provenance.

Synthetic n = 240 carries the performance numbers; a separate real-data run carries the operational-environment claim. Fair baseline throughout: RAG-with-citations + most-recent-record-wins for conflicts.

Benchmark
No runs yet. Press Run benchmark to begin.

Generates a deterministic synthetic corpus of 240 Tasks across six archetypes, computes bootstrap 95% CIs on every metric, breaks accuracy down by conflict type, and renders a separate real-data operational panel.