FICTIONAL / SYNTHETIC DEPLOYMENT MODEL. "Forge Property Management" is an invented customer. No real customer data was used and no real production results are claimed. All figures come from a deterministic synthetic corpus and stated illustrative assumptions.
FForge Outcome Delivery
100,000 Work Order Simulation

Outcome Contract

← Back to case study

Outcome Contract — Forge PM 100,000 Work Order Simulation

FICTIONAL / SYNTHETIC DEPLOYMENT MODEL. Forge Property Management is an
invented enterprise customer. No real customer data was used.

What this outcome is

A scaled, synthetic enterprise simulation that runs the **existing, verified Work Order Agent Ecosystem** (delivery-package/work-order-agents) against a deterministic corpus of 100,000 property-management work orders modeled on a fictional 42,000-unit portfolio, then quantifies operational performance and a transparent ROI versus a fully-manual baseline.

What this outcome is NOT

  • It is not a real customer deployment or case study.
  • It does not use real tenant, property, vendor, or financial data.
  • It does not claim realized production results or PRODUCTION_VALIDATED

status.

Scope (in)

  1. Synthetic dataset — 100,000 labeled work orders with unit/property/region,

trade category, priority, tenant description, timestamp, duplicate patterns, missing-field cases, emergency cases, vendor candidates, estimated cost, SLA requirement, and the expected classification/routing/validation outcome.

  1. Simulation runner — boots the real ecosystem (PostgreSQL via PGlite + gRPC

dispatch) and processes all 100,000 orders through classify → route → validate → action, measuring twelve operational metrics.

  1. ROI model — manual vs. agent-assisted labor economics from stated

assumptions, with full arithmetic.

  1. Executive report — a PDF-ready impact study.
  2. Web case study page — a public portfolio page with the synthetic disclosure.
  3. Social package — marketing copy that labels itself a fictional simulation.
  4. Proof package — dataset method, assumptions, ROI sheet, simulation results,

audit trace, checksum manifest, limitations, and verification report.

Disclosed seams

  • Fictional customer; synthetic data; simulated operational + financial assumptions.
  • Live infrastructure is in-process (PGlite + loopback gRPC); external Postgres and

a Go gRPC service are wire-compatible but not exercised here.

  • The classifier is a deterministic lexicon engine, not a hosted LLM.

Success criteria (MUST_PASS, asserted by verify.mjs)

  • All 100,000 orders processed through the live ecosystem.
  • Classification ≥ 0.90, priority ≥ 0.90, routing ≥ 0.99 on the synthetic key.
  • Exception recall ≥ 0.95; false-auto-action ≤ 0.02; precision ≥ 0.90.
  • Duplicate suppression ≥ 0.98; missing-field/over-cost/missing-location safety = 100%.
  • 100% audit completeness; dispatch idempotency on the wire; malformed rejection.
  • Deterministic corpus (seed-stable fingerprint).
  • Positive first-year net savings and a computed payback period under the model.

Certification posture

The Proof Layer assigns the authoritative certification state independently. Because the customer and data are fictional (a disclosed outcome seam), the state will not be CERTIFIED or PRODUCTION_VALIDATED; the simulation-tier label is FICTIONAL_DEPLOYMENT_MODEL_CERTIFIED.