What is real vs. modeled
We are deliberate about honesty. This platform blends live data from the application database with a deterministic business model so leadership can evaluate the operating concept without exposing production ERP systems. Here is exactly where each number comes from.
Live platform data (real)
The persona worklists, agent-swarm run traces, Value Ledger, governance ledger (audit trail, eval harness, cost meter, prompt registry), human-in-the-loop queue, and the executive KPI roll-up all read from the live application database. Approving a queue item writes a realized Value Ledger entry and a signed audit record, then the executive KPIs recompute — a real closed loop.
Deterministic simulation (modeled)
Each MVP command center (Agent Mesh, lifecycle waterfalls, 360° entity views, exception queues, operations metrics) is a deterministic model calibrated to AiSO Food Corp’s scale. The same inputs always render the same outputs — no random data — so the experience is reproducible for every stakeholder.
Live drift overlay (cosmetic)
Headline figures gently “drift” on screen at a fixed rate to convey real-time activity. Drift is a presentation effect anchored to the underlying base value; it never alters the database or the booked financials.
ROI methodology & assumptions
All ROI figures are annualized and stated net of platform cost. We separate recurring annuity benefits from one-time releases, and we attach a confidence band reflecting evidence quality. Realized-to-date reflects value already booked in the Value Ledger; the remainder is modeled pipeline.
MVP 1 — Trade Deduction
Annualized from trailing 12-month deduction volume; recovery uplift modeled at validated win-rate on disputed claims net of platform cost. Excludes one-time implementation.
MVP 2 — Collections
Cash-flow improvement is a one-time working-capital release from a 3.2-day DSO reduction (not a recurring P&L gain); recurring benefits are labor and bad-debt avoidance. Carrying-cost value of freed cash is excluded to stay conservative.
MVP 3 — Invoice Exception
Per-invoice cost reduction applied to 34K annual invoices; early-pay capture assumes sufficient liquidity to take discounts. Duplicate and penalty figures reflect trailing-year prevented losses, treated as recurring at steady state.
MVP 5 — Demand Sensing
Benefits modeled from an 11-pt WMAPE improvement across 680 SKU-DCs; waste and stockout values use current spoilage and lost-margin rates. Promotional lift capture assumes forecast accuracy holds through promo periods.
Note: MVP 4 (Source-to-Pay buyer assistance) benefits are presented in-context on its command center and are dominated by negotiated savings and cycle-time compression. One-time working-capital releases (e.g. the DSO reduction in Collections) are explicitly excluded from recurring run-rate so they are not double-counted.
Reset demo to a pristine state
Re-seeds governance, evals, and base records, and re-opens any decided human-in-the-loop items so a fresh walkthrough can be run. Captured value and the immutable audit trail are preserved — nothing financial is deleted.