A purchase order (PO) is a commercial document and the first official offer issued by a buyer to a seller indicating types, quantities, and prices for products or services.
The Purchase Order Skill consists of 2 Document Skills that are part of a processing skill in Vantage, handling regional requirements for purchase orders from the USA and Europe. The skill has been trained on a large document set from the supported regions and is capable of processing purchase orders in a variety of input formats and extracting key data points from documents in 4 languages with high accuracy. The extracted data can then be passed on to business systems and processes managing order entry, invoicing, and delivery. Processing purchase orders in Vantage can be further improved by adding a manual review and verification step.
- Receives and processes purchase orders in a variety of input formats - scanned images, PDFs, email attachments etc. Provides high accuracy of OCR conversion and data extraction.
- Includes two individual Purchase Order Document Skills with regional specifics for USA & Europe (Germany, France, Spain).
- Automatically detects the purchase order region and applies classification, routing the document to the corresponding Invoice Document Skill.
- Pre-trained model automatically detects key fields on the purchase order and extracts relevant data (buyer, order number, order date, ship to and bill to addresses, line items, etc.). For a full list of extracted fields, please see the skill documentation.
- Detects Buyer and Supplyer automatically, or allows to match against an existing database for optimal accuracy
- Includes business rules and performs automatic validations such as required fields, total amount calculation, delivery date validation etc. Further business rules can be added as individually required.
- Can be augmented by human-in-the-loop review and verification based on accuracy thresholds, missing or incorrect information. The system learns from human interaction and constantly improves the model over time.
- Allows further training of the model based on customer documents in real time. Over time significantly improves straight-through-processing rates and reduces the need for human interaction.