Document AI offers multiple products to process documents for informationfor different use cases.

Pretrained parsers

For more information, go toExplore pretrained processors.

Bank statement parser

Bank statement parser extracts key-value pairs (KVP). It can extract upto 17 generic entities. Examples include: Account number, client name, bank name,and table items like deposits and withdrawals. You don't specify the fields(schema) you want to extract. Bank statement parser supportsEnrichmentandNormalization.

W2 parser

W2 parser extracts from the IRS Form W2 as KVP. It can extract upto 12 generic entities, including employee name, Social Security Number,employer, and wages. You don't specify the fields (schema) you wantto extract. W2 parser supportsEnrichment.

US passport parser

US passport parser extracts KVP. It can extract up to seven generic entities. These include given names, family names, document ID, anddate of birth. You don't specify the fields (schema) you want toextract. US passport parser supportsNormalization.

Utility parser

Utility parser extracts KVP. It can extract up to 75 generic entitiesfrom utility bills. These include supplier name, previous paid amount, and line items like amount, description, and product code and quantity. You don't specify the fields (schema) you want to extract with theutility parser.

Identity document proofing parser

Identity document proofing parser predicts the validity of ID documentsusing multiple signals.

• fraud_signals_is_identity_document detection: Predicts whether an image contains a recognized identity document.
• fraud_signals_suspicious_words detection: Predicts whether words are present that aren't typical on IDs.
• fraud_signals_image_manipulation detection: Predicts whether the image was altered or tampered with an image editing tool.
• fraud_signals_online_duplicate detection: Predicts whether the image can be found online (US only).

Pay slip parser

Pay slip parser extracts KVP. It can extract up to 26 generic entities from payslips. These include employee name, bonus, commissions, overtime, and pay date.You don't specify the fields (schema) you want to extract. Pay slip parser supportsEnrichment andNormalization.

US driver license parser

US driver license parser extracts KVP. It can extract up to eight generic entitiesfrom a driver license. Examples include: Given name, family name, document ID, andexpiration date. You don't specify the fields (schema) you want toextract. US driver license parser supportsNormalization.

Expense parser

Expense parser extracts KVP. It can extract up to 17 generic entities from expensereports. Examples include: Expense date, supplier name, total amount, and currency.You don't specify the fields (schema) you want to extract. Expense parser supportsEnrichment andNormalization.

Invoice Parser

Invoice Parser extracts KVP. It can extract up to 46 generic entitiesfrom invoices. These include invoice number, supplier name, invoice amount, taxamount, invoice date, and due date. You don't specify the fields (schema) you want to extract. Invoice Parser supportsEnrichmentandNormalization.

Summarizer

Summarizer gives abstract and bullet pointsummaries for short and long documents. Summarizer also lets you specify the output length of the summary as comprehensive, medium, or brief.