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

Run & Deploy

← Back to case study

Run & Deploy — Forge PM Work Order Simulation

FICTIONAL / SYNTHETIC DEPLOYMENT MODEL. No real customer data is used. This
document describes how to run the *simulation*, and how a real deployment of the
underlying ecosystem would differ.

Run the simulation locally

cd delivery-package/forge-pm-work-order-sim
node verify.mjs            # full 100k run (~15-20 min)
node build-deliverables.mjs
node publish.mjs

Proof + gates (from repo root)

node tools/forge-proof.mjs --outcome delivery-package/forge-pm-work-order-sim \
  --verify "node verify.mjs" --report verification-report.json \
  --outcome-seam "Fictional enterprise customer (Forge Property Management) and 100% synthetic data/assumptions; no real deployment."

node tools/forge-proof-verify.mjs --outcome delivery-package/forge-pm-work-order-sim
node tools/forge-gate.mjs --outcome delivery-package/forge-pm-work-order-sim --action publish
node tools/forge-portfolio.mjs

What a *real* deployment would require (disclosed seams → live)

This simulation runs the ecosystem with in-process infrastructure and synthetic input. A real deployment for an actual property manager would additionally need:

  1. Real intake integration. Replace the synthetic generator with the customer's

work-order intake (portal, email, phone transcription, IoT) feeding the same order shape.

  1. External database. Set DATABASE_URL to a managed PostgreSQL (the repository

and gRPC service use the same SQL via the pg adapter). PGlite is the in-process engine used here.

  1. External dispatch service. Point DISPATCH_GRPC_URL at a deployed dispatch

service (a Go implementation of proto/dispatch.proto is wire-compatible) and integrate the customer's vendor/dispatch systems.

  1. LLM classifier (optional). Swap the deterministic lexicon classifier for an

LLM behind the same interface, and re-verify accuracy on the customer's real text.

  1. Identity, RBAC, tenant isolation, retention. Enterprise non-functional

controls, evaluated by the Proof Layer before any production claim.

  1. Live validation. Only real production evidence + an official benchmark +

independent reproduction could ever justify a PRODUCTION_VALIDATED state — which this synthetic model explicitly does not claim.

Determinism

Everything is seeded (20260625). Re-running the generator reproduces the exact corpus (verified by the corpus fingerprint check), so all reported numbers are reproducible by any independent engineer.