Skip to content

Stop paying people to re-type documents.

IntELIEDocs works out what arrived, pulls the fields and line items your systems need, and keeps the page coordinates behind every value — so a disputed number is settled by looking at the page, not by arguing.

  • Your own Smart Classes
  • Fields, tables & line items
  • Human review gate
EliehubIntELIEDocsInvoice Mapper
Apple-InvoicesInvoice_4471.pdfProcessingCompleted
Stanford Plumbing & Heating
123 Madison drive, Seattle, WA 7829Q
www.plumbingstanford.com

BILL TO

Allen Smith
DESCRIPTIONQTYTOTAL
Installed new kitchen sink3150.00
Toto sink1500.00
Worcester magnetic filter1190.00
Nest smart thermostat1250.00
ResultDetailsGroups

BASIC INFORMATION

Vendor Name
Stanford Plumbing & Heating
Vendor Address
123 Madison drive, Seattle, WA 7829Q
Vendor Website
plumbingstanford.com
Bill To
Allen Smith
LINE ITEMS4 found

What it does

Paper in, keyed data out.

Works out what arrived

Every document is matched to a Smart Class you defined, with a confidence score attached to the decision. Nothing is forced into a taxonomy someone else picked.

Pulls fields, tables and line items

Repeating rows come back as rows, not as a wall of text, and each value keeps the bounding box it was read from so you can always see the source.

Asks a human when it should

Per-field confidence decides what needs a person. Low-confidence results route to review, and nothing partial is published downstream.

Use cases

Where teams point it first.

  1. 01

    Accounts payable keys supplier invoices into the finance system by hand.

    Invoices arrive as typed fields and GL-ready line items; the team reviews exceptions instead of the batch.

  2. 02

    Onboarding packs turn up as one mixed PDF with no consistent order.

    Each document is split out and classified, and the missing one is named rather than discovered weeks later.

  3. 03

    Someone disputes a total that was transcribed from a scan months ago.

    The stored page coordinates put the original value back on screen, so the question is settled by looking rather than arguing.

  4. 04

    Volume spikes and the queue is handled by whoever is on shift.

    The same class and schema run every time, so output does not change with the person processing it.

How to use it

Four steps from first file to delivered data.

  1. /intelie-docs/smart-classes/create

    Define a Smart Class

    Describe a document type you actually receive and the fields you need from it. Set the visibility scope now so the right team sees the results later.

  2. /smart-extract/create · /views

    Map it to your schema

    A Smart Extract Mapper names each field the way your downstream system does, and views let different teams see the same extraction in their own shape.

  3. /intelie-docs/upload · import-channels

    Send documents in

    Drop files in the workspace to start, then switch to an import connector once you trust it. Re-sends are de-duplicated on content, so a replayed mailbox costs you nothing.

  4. /$documentId · export-channels

    Review the exceptions, then deliver

    Open the workbench, click a field, see the box it came from and correct it. Mark as reviewed, and delivery fires — either on your send or automatically for connector-driven work.

Bring us your ugliest document.

Send the one nobody wants to open. We'll run it through the real pipeline and show you the fields, the line items and the point where it would have asked a human.