What VectorLedger Solves
Two pre-existing datasets, one decisive join
Drawings sit in scanned PDFs, PLM and network drives where nobody can search them. The ERP ledger is searchable but blind to what a part actually is. The same part carries different identifiers across drawing, item master, invoice and second plant, so near-identical parts bought at different prices can go unnoticed.
Capability Flow
From scanned drawing to evidenced finding
Six stages convert unreadable archives and raw ledgers into findings a sourcing professional can defend.
Read the archive
Scan clean-up, deskew, OCR and title-block extraction using templates learned from about twenty labelled examples.
Normalise purchases
PO and GRN lines, vendor identity and price series. Every excluded line keeps a reason.
Resolve identity
A seven-strategy cascade matches drawing, part and PO, biased against false merges.
Cluster families
Text, visual, dimensional and specification similarity group parts into families.
Normalise scope
Eleven commercial dimensions, each marked known, inferred or unknown, with provenance.
Compute findings
Variance, an adjustment ledger, addressable exposure, a confidence grade and mandatory limitations.
Pipeline stages as implemented in the VectorLedger engine.
Value Chain
Five stages that are never summed or conflated
VectorLedger claims identification only. Each stage after that is owned by a named person.
Observed variance
The raw price difference across vendors for the same or equivalent part.
Addressable exposure
Observed variance minus an evidenced adjustment ledger for scope differences.
Validated opportunity
A sourcing professional agrees the finding is real.
Negotiated saving
A named owner agrees a new price with the vendor.
Realised saving
Confirmed by the customer's finance team after the price takes effect.
Product Map
From the opportunity board to the source document
Every screen exists to make a number inspectable.
Capabilities
What VectorLedger does for sourcing and engineering
Each capability is a module in the engine.
Searchable drawings
Ingests, scans for malware, hashes, deduplicates and renders originals, then reads title blocks with calibrated confidence and a per-customer template.
- Includes: Works on scanned PDFs, not only clean CAD exports.
- Includes: Extraction confidence is reported, not hidden.
One part, one identity
Resolves the same part across drawing numbers, ERP codes, invoice text and second-plant IDs, with a bias against false merges.
- Includes: Variants such as AX-00123-R4, 00000000123 and AX123 are designed to resolve together.
- Includes: Unresolved identities stay visible instead of being forced.
Comparable vendor prices
Imports PO and GRN lines, resolves vendors and builds price series, keeping a reason for every excluded line.
- Includes: Same-part multi-vendor and single-source findings.
- Includes: Unexplained price increases and equivalent price bands.
Find equivalent parts
Combines text, visual, dimensional and specification similarity to cluster near-identical parts into families.
- Includes: Surface duplicates that carry different part numbers.
- Includes: Cluster explorer is on the product roadmap.
Compare like with like
Eleven dimensions: material supply, treatment, freight, packaging, tooling, inspection, certification, payment terms, MOQ, lead time and programme context.
- Includes: Each dimension is known, inferred or unknown.
- Includes: Adjustments are itemised in a ledger.
Findings with confidence grades
Each finding carries a confidence grade and mandatory limitations. Findings below the evidence threshold are withheld.
- Includes: Opportunity board has no sort by raw variance on purpose.
- Includes: Annotations, saved views and exports for the team.
Trust and Governance
Principles built into the product
The product principles are enforced in the arithmetic and the interface.
Deterministic arithmetic
Money is computed with integer paise arithmetic. Models help read drawings and find similarity, not produce prices.
Provenance everywhere
Every number resolves to a source drawing region or an ERP row within three clicks.
Human owns the claim
The product identifies exposure. Validation, negotiation and realisation belong to named people.
No cross-tenant pooling
A customer's drawings and prices are never pooled with another customer's data.
Who Uses It
Built for sourcing, engineering and finance together
Designed for machinery manufacturers with large drawing archives and bought-out spend.
Sourcing managers and CPOs
Review vendor and part findings with evidence and scope adjustments.
Design and cost engineers
Find existing parts and tooling before releasing a new design.
Data stewards
Resolve unmatched identities and maintain title-block templates.
Finance
Confirm realised savings against agreed prices.
Frequently Asked Questions
VectorLedger: common questions
What does VectorLedger do?
VectorLedger makes engineering drawings searchable and part prices comparable. It joins the drawing archive to the ERP purchase ledger and produces auditable price-variance findings adjusted for commercial scope.
Does VectorLedger use AI?
VectorLedger uses OCR and similarity models to read drawings and group parts. All money arithmetic is deterministic, and findings without enough evidence are withheld instead of estimated.
Which ERP systems does it work with?
VectorLedger starts from purchase-ledger and PO/GRN exports such as CSV from SAP, Oracle, Infor or Tally. Live ERP connectors are on the roadmap.
Does VectorLedger promise savings?
No. It identifies addressable exposure and shows its working. Validation, negotiation and realisation are tracked as separate stages owned by named people.
How does an engagement start?
A time-boxed Spend Discovery pilot runs the pipeline on a customer's own archive and ledger to test whether defensible exposure exists.
Explore More
Related Mativyx products
See how VectorLedger sits alongside the rest of the Mativyx platform family on the products overview.
Flow
Tender, DPR, audit and cable-sizing workspace for EPC and industrial teams.
Coherence
Multi-plant industrial operations: compliance, permits, quality, contracts and OEE.
Tessera
Retail shelf intelligence: live shelf state, alerts and replenishment proof.
Talk To Mativyx
Find out what your drawings and ledger already know
Start with a Spend Discovery pilot on your own drawing archive and purchase ledger.
Request Spend Discovery