Veryfi AI is trained to extract out-of-the-box a wealth of fields from various types of documents in seconds using machines end-to-end. For you, this means you will be up and running much faster than alternative solutions on the market.
Veryfi processing runs these 3 steps:
1. OCR your document
2. Data Extraction using a pre-trained machine model to give context to your document and turn it into structured data
3. Data Transformation categorizing the document, checking for duplicates and executing post-processing & enrichment steps
Data Enrichment
Vendor Details Expansion
Veryfi can identify vendor name by the just logo in the case when Vendor name is not listed on the document
Veryfi can identify vendor by line items. If there is no Vendor name and no Vendor logo on the document, Veryfi will look into line items on the receipts/invoice and will try to find Vendor based on a specific product. For example, it will identify BurgerKing if the receipt has a “whopper sandwich” and other items relevant to just BurgerKing
In the API response, Veryfi provides 2 vendor name fields “raw name” and “name”. Raw name is the way this vendor name is spelled on the document, Name is the confirmed vendor name against our database of vendors. This name can come from matching by phone number, address, VAT id, or other means
Veryfi also has a database of Vendor logos that are returned in the response, those logos can be used to build a nice UI/UX
Vendor address can be expanded, validate, and parsed into an address object (address line, city, state, zip, etc) from a short address that is listed on a document
Read more about Vendors
Document Categorization
Veryfi can categorize documents using client’s GL accounts. When partner submits a document for data extraction to Veryfi API, a partner can also send a list of GL accounts in the same request and when Veryfi processes this document it will choose the best category from that list of GL accounts
Read more about Categorisation
Worldwide Currency Detection
Veryfi can identify a currency on the document even if the currency is not listed on the document. Veryfi’s model will use a lot of parameters like address, tax codes business registration number to correctly identify the currency.
List of Currencies Veryfi support
Line Items Expansion & Categorization
Veryfi expands shortened line items descriptions, ie “Blbry Ygrt” will be expanded to “Blueberry Yogurt”
Veryfi also assigns a category/type to each line on the receipt and invoice, it can also use the client’s GL accounts to categorize each line item
Line item types available: room, tax, parking, service, fee, delivery, product, food, alcohol, tobacco, transportation, fuel, refund, discount, payment, giftcard . More in API Integration guide
Date format Identification
If the Date on the invoice is in MM/DD or DD/MM format Veryfi will predict the format of the date on the document, this is very important to Canadian receipts and invoices since in Canada both format DD/MM and MM/DD are valid
Read more about Date field
Due Date Calculation
If Due Date is not listed on the invoice and payment terms are provided (NET30, Due on Receipt, etc) Veryfi will calculate the Due Date using payment terms and invoice date
Reading Barcodes / QR codes
Sometimes important information like Date, store ID, or Payment URL is encoded into Barcodes (EAN / UPC), QR codes, or PDF417. Veryfi reads and decodes those codes
Read more about Barcodes
⏰ Compute flag disables some calculations and allows the developer to infer their own results. By default, it is always true unless a user disables computation by setting the flag to false. If a user specifies compute:false the fields will not be enriched on Veryfi side. Read more here.
🧝🏻 How do I know if a field was enriched on Veryfi side? If you use confidence details, you can tell it by missing scores and bounding boxes for such values. Read more
Other Related Articles:
Please, reach out to support@veryfi.com if you have any questions.
Veryfi tech specialists will be happy to discuss your use case more in detail.