ZAAP · Capabilities · v1.0
Three phases.No skipping gates.
EVALUATE forces every claim through a lie detector. PILOT runs the proposal under a fixed ceiling. GOVERN keeps production honest with a WORM ledger and codified kill-switch criteria.
01 — EVALUATE
EVALUATE. The Lie Detector.
Before any dollar is committed.
Every proposal is reconciled against the customer's actual process data before approval. Claimed ROI is published next to ZAAP-validated ROI with an explicit error band.
- 01ROI Error Band — claimed vs. validated hours, savings, and payback
- 02As-Designed VSM extracted from policy and SOPs
- 03As-Executed process map via PM4Py on real event logs
- 04Compliance surface auto-mapped at intake (DoDFMR, ARC-AMPE, EU AI Act, ISO 42001)
- 05Archetype shortlist and risk tier proposed from intake answers
- 06Reviewer Gate 2D — As-Designed evaluation outcome and rationale
- 07Self-critique advisories surfaced before submission
Workflow
- 01Intakestep 1
- 02As-Designed VSMstep 2
- 03As-Executed PM4Pystep 3
- 04ROI Error Bandstep 4
- 05Gate 2Dgate
- 01
Intake
step 1
- 02
As-Designed VSM
step 2
- 03
As-Executed PM4Py
step 3
- 04
ROI Error Band
step 4
- 05
Gate 2D
gate
02 — PILOT
PILOT. The Sandbox.
8-week governed sandbox under a fixed ceiling.
An 8-week governed pilot under a fixed sub-$500K ceiling. Scope, evidence, and exit criteria are locked at gate entry. Conditional action items are tracked to closure.
- 01Fixed scope, fixed ceiling — no scope creep, no budget drift
- 02Air-Gapped OCR keeps source documents inside the customer enclave
- 03Pydantic Lock rejects hallucinated fields at the boundary
- 04Conditional Action Items with owner, due date, and aging
- 05Evidence capture with sha256 on every artifact
- 06Reviewer Gate 3C — As-Executed conformance and ROI update
Workflow
- 01Pilot Charterstep 1
- 02Air-Gapped OCRstep 2
- 03Conditional Actionsstep 3
- 04Evidence Capturestep 4
- 05Gate 3Cgate
- 01
Pilot Charter
step 1
- 02
Air-Gapped OCR
step 2
- 03
Conditional Actions
step 3
- 04
Evidence Capture
step 4
- 05
Gate 3C
gate
03 — GOVERN
GOVERN. The Ledger.
Production lives on the ledger, not the slide deck.
Production deployments live on a WORM audit ledger. Drift, bias, and incident triggers force structured re-review. Kill-switch criteria are codified, not implied.
- 01WORM, append-only audit ledger across every artifact and decision
- 02Bias and drift monitoring via Fairlearn-style gates
- 03Incident heat with severity classification (Sev1–Sev4)
- 04Re-Review Triggers — codified, not discretionary
- 05Benefits realized vs. claimed reconciliation per period
- 06Kill-switch criteria recorded at approval, enforced in production
- 07Reviewer Gate 4D — final disposition and rationale
Workflow
- 01Productionstep 1
- 02Drift + Biasstep 2
- 03Incident Heatstep 3
- 04Re-Review Triggerstep 4
- 05Gate 4Dgate
- 01
Production
step 1
- 02
Drift + Bias
step 2
- 03
Incident Heat
step 3
- 04
Re-Review Trigger
step 4
- 05
Gate 4D
gate
04 — Comparison
ZAAP vs. the alternatives.
Categories, not vendors. No logos by design.
| Capability | ZAAP | Generic AI Platform | PPM / EPMO Suite | RPA / iPaaS Tool |
|---|---|---|---|---|
| Claim vs. as-executed ROI reconciliation | Yes | No | No | No |
| As-Designed VSM extracted from policy | Yes | No | Partial | No |
| As-Executed process mining (PM4Py) | Yes | No | No | Partial |
| Compliance surface mapped at intake | Yes | Partial | Partial | No |
| Air-gapped OCR inside customer enclave | Yes | No | No | No |
| WORM audit ledger with sha256 evidence | Yes | No | Partial | No |
| Bias and drift gates with re-review | Yes | Partial | No | No |
| Vendor-agnostic referee posture | Yes | No | Yes | Partial |
Categories, not vendors — no logos by design.
Want to see the lie detector run against your own claim?
Schedule Workshop