STATIC PAGE Interactive levers, live recompute, and Record decision run in the app build; every figure, formula, evidence row, and WHY below is the real render, pre-computed. Evidence panels are opened for review.
Each specialist is a deterministic engine over its domain modules: every figure recomputes from source, every answer carries evidence and lineage, and a claim whose inputs cannot be reproduced is withdrawn rather than guessed. The model-routed layer: provider decided, a selected frontier model provider (D-R52), AT_ACTIVATION until deploy credentials let a live call fail red; the deterministic specialists answer until then.
Assistant console
Deterministic assistant router over the tenant engines; generative layer AT_ACTIVATION. Ask anything the engines can prove; anything else refuses with a code and a safe alternative.
Bring whatever model you trust. We will make the organization trust the decision.
Model-agnostic by architecture, not by policy. Every recommendation arrives with its evidence, calculation, assumptions, confidence, and validation, and that contract is enforced above whichever model produced it. Swap the model and the contract is unchanged, which is what makes the choice yours to make and ours to survive.
- It does not claim a better model than the one you already trust.
- It does not write to your systems of record, so a disagreement about a figure is never also an incident.
- It does not decide. It records who decided, on what evidence, under what authority, and what happened next.
- It does not claim analyst placement, a reference customer, or a live deployment. It claims an artifact you can verify yourself, without asking us.
Read-only by architecture. Answers are MODELED on SYNTHETIC records and carry their lineage; refusals carry their code.
why do we have this inventory on GC-4471
Cause 1 via supplies: Supplier terms set the batch: MOQ 5000 from Vendor-Alpha Forging. Cause 2 via supplies: Lead time of 6 weeks forces protection while replenishment travels. Cause 3 via demands: Customer demand pulls 1400 units a week with variability sigma 240. Cause 4 via drives: Cycle stock and safety stock together are the inventory position.
5600 units per order; 967 units of safety stock; 9367 unit reorder point; 3767 average units, 155200 USD MODELED
orderQty = max(MOQ 5000, demand 1400/wk x cover 4w) = 5600 | safetyStock = z 1.645 x sigma 240 x sqrt(LT 6w) = 967 | ROP = demand 1400 x LT 6 + SS 967 = 9367 | avgInv = Q/2 (2800) + SS (967) = 3767 units
Sourced edges only; SYNTHETIC records, MODELED figures.
sourceSystem: SAP S/4HANA · sourceId: DET-01/MAT-4471
what if MP-5120 moq to 6000
The lit path runs moq through orderQty, avgInventoryUnits, inventoryValue, daysOfSupply.
Baseline vs scenario across 13 metrics; 8 nodes on the affected path.
orderQty = max(MOQ 6000, demand 3800/wk x cover 3w) = 11400 | safetyStock = z 1.645 x sigma 610 x sqrt(LT 3w) = 1738 | ROP = demand 3800 x LT 3 + SS 1738 = 13138
Safety stock recalculates with lead time, holding the service target constant; the cost of a longer lead time shows up as working capital, not as fill-rate risk.
sourceSystem: NetSuite · sourceId: STG-02/ITEM-5120
enterprise health
Margin 100, turns 100, integrity 80, service 96.
Revenue $13,402,480 MODELED; WC $469,198 MODELED; contested 1.
composite = mean(margin 100, turns 100, integrity 80, service 96) = 94
Composite is a plain mean; every component recomputable.
sourceSystem: EOS model · sourceId: healthOverview/full-year
This layer is read-only by architecture: it recommends and explains; execution happens in your systems of record.
Safe alternatives: ask why a figure is what it is · run a what-if in the impact simulator · record a decision on the ledger with your authority