Methodology — how the toolkit reads, scores and verifies

One promise across every module: every figure in every output traces back to a source line or a clause — and a human verifies it before it becomes evidence.

What happens to your upload

read · sort · extract · verify

Step 01

Read & sort

Every file in the bundle is read — PDFs, scans, workbook tabs. Anything that isn’t a maintenance document (a photo, a personal note) is set aside with a reason, never silently used.

Step 02

One common standard

Contracts, invoices, workscopes, task cards and lease terms are read into one common standard, so spend findings and contract findings reconcile against the same evidence.

Step 03

Diagnose & score

The module’s diagnostics run on that clean base and score it against our published standards — spend against billed cost, contract architecture, redelivery readiness, engine condition.

Step 04

Verify & gate

Free Tier 01 output is indicative and watermarked. From the Stage 1(a) data pack up, a human verifies line-level evidence before anything is relied on.

Bundle pipeline: read, sort, extract, verify — source documents are not stored, they are processed transiently and discarded

What we score against

published, versioned standards — not black boxes

Cost intelligence & contract architecture

13 spend generators · 19 scored contract sections

A spend audit against billed cost, and a contract-architecture score against our maintenance-contract standard — the same standard used in Tier-03 RFP work.

End-of-Lease contract review

219 line items · 13 return-condition systems

Every return condition and compensation rate extracted and scored for redelivery readiness, with settlement exposure modelled on a half-life basis. Lease methodology →

Engine evaluation

Technical · records · market subscores

LLP stack and shop-visit history assessed row by row into a GO / CAUTION / NO-GO verdict with a risk rating on each subscore.

Versioning

Versioned methodology (v0.3.0)

Standards are versioned; a deliverable states the version it was scored against, so results are comparable across time and assets.

Evidence discipline

what makes the output usable in a negotiation

Traceability. Every figure, score and flag carries its source — the document, the line or the clause it came from. Nothing is asserted that can’t be pointed at.
Human verification. AI drafts, people confirm. From Stage 1(a) up, extracted evidence is verified by a reviewer before it is relied on; expert-led review from Tier 02.
Document handling. Tier-01 diagnostics don’t store your pasted documents — results are computed and the source discarded. Retention above that is explicit, consent-gated and time-boxed. See data handling.
Honest posture. Free-tier output is labelled indicative. Verified means a human has checked it — the tier ladder is the verification ladder.

The same standards run from the free diagnostics to Tier-03 engagements — the tiers change how much evidence is verified and who works it, not the method. How it works Engagement levels