AIDA
11/5/2024

Intelligent document processing (IDP): meaning, how it works

Intelligent document processing (IDP): meaning, how it works

Intelligent document processing (IDP) is the technology that reads the documents a business receives, recognises their type, extracts their data and checks it before passing it to the ERP or the archive. Where OCR stops at the text, IDP understands what it means: it knows that a number is the total of an invoice and that a date is the expiry of a contract.

AIDA is an IDP platform that works this way on invoices, contracts, delivery notes, orders, forms and correspondence, without training: it recognises dates, VAT numbers and IBANs straight away, learns a new field with one click on an example and, with AIDA Boost, extracts every field from the very first document. In 2026 AIDA was included in the Infosource Global Capture & IDP Software Vendor Matrix 2026.

How does intelligent document processing work?

An IDP system takes every document from the moment it reaches the business all the way to the ERP, in five steps that, with AIDA, happen without manual intervention when the data is reliable.

  1. Capture (document capture): documents arrive by email, scanner, shared folders, smartphone or WhatsApp, including in mixed batches or in a single PDF containing many of them.
  2. Classification: the system splits the documents contained in the same file and recognises the type of each one, for example an invoice, a contract or a delivery note.
  3. Data extraction: OCR reads the text, handwriting included, and IDP finds the fields you need, such as supplier, number, date, amounts, IBAN or the rows of a table that runs over several pages.
  4. Validation: every value is checked before it is used, and only doubtful values are passed to a person.
  5. Integration: the data and the document reach the ERP, for example SAP or Sage, and the archive, whether that is AIDA's own, with unlimited storage on every plan, or SharePoint, Google Drive or Dropbox, with a file name and folder set by your rules.

The result is document process automation (IDP automation): invoices reach accounts already posted and contracts reach the archive already renamed, without anyone typing in a single value. All the destinations are listed on the AIDA integrations page.

What is the difference between OCR and IDP?

OCR turns the image of a document into text; IDP uses that text to work out what the document is, what data it contains and whether that data is correct. OCR is therefore the first step of IDP, not an alternative to it: as we explain in our guide to what OCR is, character recognition on its own cannot tell a total from a phone number.

What changes OCR IDP
Output Searchable text Structured, validated data
Document type Not recognised Classified: invoice, contract, delivery note, form
Fields such as total, due date or IBAN Not distinguished from the rest of the text Found and converted into dates and numbers
Value checks None Comparison with already validated data and flagging of anomalies
After reading A person copies the data by hand The data reaches the ERP and the document the archive

How does an IDP system reduce document errors?

An IDP system reduces errors because it removes manual keying and checks every value before it reaches the ERP. When extracting data from invoices and contracts, where a wrong amount or date can be costly, AIDA applies several checks.

  • Lookup: it compares the values read with lists you have already validated, such as your supplier master data, normalises them and fills in missing data, for example the company name from the VAT number.
  • Anomaly detection: if a utility bill that is usually £100 shows £1,250, or a VAT number is missing a digit, AIDA flags the value and does not validate the document automatically until someone has checked it.
  • Duplicate documents: AIDA recognises duplicates from their content or from the extracted data and, if you choose, blocks their validation, so the same invoice is never posted twice.
  • Human in the loop: documents with reliable data go through on their own, doubtful ones go to whoever needs to review them, and every one-click correction also applies to the documents that follow.

All the checks are described on the AIDA AI data extraction page.

What sets AIDA apart from other IDP platforms?

Many IDP systems require you to build a template for every layout or to collect samples to train the model, and they are often designed for a single process, such as supplier invoices. AIDA was created in 2012 precisely to do away with all of this, as described in the article on template-free data extraction.

  • No training: dates, VAT numbers, IBANs, email addresses and company names are recognised straight away. A new layout is learnt when you confirm the first document and is then recognised automatically, along with similar layouts, without drawing or selecting anything.
  • AIDA Boost: it extracts every field from the first document, including those specific to your business, and also reads unstructured documents such as letters and legal deeds.
  • Complex documents: AIDA reads printed and handwritten text, rotated documents, barcodes and tables that run over several pages, in Italian, English, French, German, Spanish and Portuguese.
  • Any document, any industry: AIDA is a horizontal platform. You start with today's problem, for example supplier invoices, and use it just as easily for contracts, case files, forms or delivery notes, backed by thirty years of experience in document management.
  • From data to decisions: the Overview turns the extracted data into charts, KPIs and forecasts based on historical data, updated in real time as documents come in.

Intelligent document processing FAQs

What does IDP stand for?

IDP stands for intelligent document processing. In IT the same acronym also refers to an identity provider, the service that manages user access, which has nothing to do with documents.

Does IDP replace OCR?

No, it includes it. OCR remains the first step, the one that reads the text; IDP adds classification, field extraction, validation checks and the delivery of data to business systems.

Does an IDP system need training?

It depends on the system: many require templates or samples for every document type. AIDA does not: it recognises the most common data straight away, learns a new field with one click and, with AIDA Boost, extracts every field from the first document.

Can an IDP system read handwriting?

Yes, if its OCR recognises handwritten text. AIDA reads both printed and handwritten text, so it also extracts data from forms filled in by hand, signed delivery notes and mixed documents.

Which documents can be processed with IDP?

With a horizontal platform like AIDA, any document a business receives: invoices, contracts, orders, delivery notes, forms, letters and deeds. Every document type is recognised automatically, even when it arrives together with others in the same file.

Giorgio - CEO
Giorgio
CEO
 Go back