AI document intake — one channel for every inbound record.
Suppliers, customers and labs send documents in every format imaginable. V5 accepts them all — email attachment, portal upload, EDI drop — extracts the structured data, matches to the open PO, SO or lot, and posts the transaction with a full audit trail.
What breaks without this.
Every document type has its own inbox
PO acknowledgements, CoAs, invoices, packing lists — each channel has its own clerk, its own error rate, its own backlog.
CoAs delay QC release
Supplier CoAs arrive as PDFs that a QC analyst rekeys into the receiving inspection record. Release waits on typing.
Invoice matching is manual
3-way match (PO / receipt / invoice) is a spreadsheet reconciliation. Payment delays and duplicate payments follow.
Records-by-execution. Compliance, by design.
One inbox, every format
PDF, image, EDI 850/856/810, XML, structured email — one intake with document classification on arrival.
Extract, then match
Extracted fields are matched to the open PO, SO, lot or invoice — no separate step for the clerk to key in a reference.
CoA → receiving inspection
CoA test results feed the receiving inspection record; specs are compared server-side; passing lots auto-release to inventory.
3-way invoice match
PO, receipt and invoice matched with tolerance rules; only exceptions route to AP for review.
Full source retention
Original document retained; every extracted field carries provenance to the source region for audit and dispute.
What to look for when you're buying.
Universal intake succeeds when the extractor is layout-agnostic and the matcher is honest about exceptions. Criteria that separate the two.
Layout-agnostic extraction
What it tests: Does it need a per-supplier template or does it learn on the first sample?
Why it matters: Template-OCR breaks on every supplier tweak.
V5: Layout-agnostic; adapts on first sample.
Matching to live records
What it tests: Does the extractor match against your live PO / SO / lot data, or leave matching to the clerk?
Why it matters: The value is the match, not the extraction.
V5: Extraction and matching are one step.
Exception routing
What it tests: Are exceptions routed to the right team by exception type?
Why it matters: A price mismatch is an AP problem; a quantity mismatch is a receiving problem.
V5: Exception type drives routing.
Source retention
What it tests: Is the original document retained and linked to every extracted field?
Why it matters: Disputes and audits need the source, not the transcript.
V5: Source retained and linked field-level.
Multi-lot / multi-line handling
What it tests: Are multi-lot CoAs and multi-line invoices split correctly?
Why it matters: Composite documents are the norm, not the exception.
V5: Split at ingestion, each line matched independently.
Straight-through processing rate
What it tests: What fraction of clean documents post without human touch?
Why it matters: This is the real productivity number.
V5: 70-90% typical after first-month tuning.
Spreadsheet vs legacy QMS vs V5.
Universal intake vs a data-entry team and vs template-OCR tools.
| Capability | Spreadsheet | Legacy QMS | V5 Ultimate |
|---|---|---|---|
| Layout independence | N/A | No (per-template) | Yes |
| Live matching | Manual | Separate step | Native |
| Exception routing | None | Single queue | By exception type |
| Source retention | Filed in email | Attached to record | Linked field-level |
| Multi-lot CoA | Manual split | Sometimes | Native split |
The clauses, verbatim — and how V5 answers each.
Where regulations touch inbound-document handling.
Each lot of components... shall be examined visually for appropriate labeling as to contents and, when needed, tested to determine identity and specifications.
V5: CoA feeds receiving inspection with spec comparison at ingestion; QC has evidence to release without re-typing.
Each manufacturer shall establish and maintain procedures for acceptance of incoming product.
V5: Receiving acceptance record built from CoA + inspection + PO with full source retention.
Internal control over financial reporting.
V5: 3-way match is systemic; overrides require attestation.
Records must be provided within 24 hours of a request.
V5: Traceability records including lot-level documents are queryable and exportable in seconds.
Step by step on the floor.
The universal intake goes live in four weeks with modest tuning.
- 1Week 1
Channels connected
Email intake, portal upload and EDI drop wired to the intake bus.
- 2Week 2
Master data anchored
PO / SO / lot / supplier data anchored so matches can happen.
- 3Week 3
Exception routes set
Exception type → owner routes defined.
- 4Week 4
Straight-through on
Auto-post enabled for clean documents; exceptions queued.
- 5Ongoing
Tune thresholds
Auto-post rate reviewed monthly; thresholds tuned per document type.
The math, with the assumptions visible.
ROI is data-entry avoided, faster QC release and cleaner AP.
Data-entry FTE on inbound documents
Human only touches exceptions.
CoA-to-release cycle
Spec comparison at ingestion; auto-release when clean.
Duplicate invoice payments
3-way match is systemic, not spot-check.
Payback is typically inside 6 months on labour and duplicate-payment avoidance alone.
What changed on the floor.
Setting
A regional food manufacturer with 4 sites and 200+ active suppliers.
Before
Two data-entry clerks per site; 2-day median CoA-to-release; occasional duplicate invoice payment.
After
Six months post-cutover, clerk workload halved and reassigned to supplier-improvement work; median CoA-to-release under 30 minutes; zero duplicate payments in the last two quarters.
Proof points
- Sub-minute median from arrival to matched transaction
- Human touch only on exceptions — clean documents auto-post
- Original PDF retained and linked to every extracted field
- One dashboard for AP, QC and receiving intake status
Built to satisfy
- 21 CFR 211.84 (testing and approval of components)
- 21 CFR 820.80 (receiving, in-process and finished device acceptance)
- 21 CFR Part 11 (electronic records)
- FSMA 204 (traceability documents)
- SOX §404 (invoice controls, for public filers)
Related on V5
Frequently asked questions
Does it work with supplier CoAs that change format every quarter?+
Yes — extraction is layout-agnostic. Once a new format arrives, the model adapts on the first sample; confidence scores drive whether the clerk reviews or the record posts directly.
How is this different from RPA / template OCR?+
Template OCR breaks the moment a supplier tweaks a header. V5 uses layout-agnostic extraction and matches against your live PO/SO/lot data — the extractor doesn't need a per-supplier template.
Does it handle multi-lot CoAs?+
Yes — a single CoA covering multiple lots is split and matched per lot, each with its own inspection record.
See V5 on your own line.
Free trial, no card. Live in 7 days with guided onboarding.