The Quantis methodology — 8-layer deterministic pipeline
Quantis calculates numbers deterministically before AI ever runs. AI writes the narrative around those numbers; it does not invent them. Layer 8 (RVCE) cross-validates the model against live market data before the report is returned.
The 8-layer deterministic pipeline
- Layer 1 — InputSanitizer plus NAICS classification of the described business
- Layer 2 — BenchmarkMatcher across 97,500 real industry data points, 46 sectors
- Layer 3 — CityMultiplierEngine adjusts for 51+ cities' cost bases
- Layer 4 — DeterministicFinancialEngine calculates 12 financial variables through 12 quality gates
- Layer 5 — MonteCarloEngine runs 1,000 iterations with P10/P50/P90 percentile projections
- Layer 6 — Edge Function model layer (classify, narrative, action plan, adversarial review)
- Layer 7 — SourceAttributor tags every number with the benchmark row it derived from
- Layer 8 — Reality Validation and Calibration Engine (RVCE): cross-checks projections against live market data
See the research · Security posture