Document Automation processing workflow

Document Automation takes data in from a large range of formats and uses multiple processing steps and various technologies to convert that data into structured, usable information. This structured information is then typically used in downstream steps for updating systems of records, making decisions, or further processing aggregated data.

Document Automation uses AI technologies for processing documents in the following phases:

Different phases of processing documents in Document Automation

Document ingestion
This is the first step in document processing where documents are collected and imported from various sources such as email attachments, file systems, Cloud storage, APIs, and content management systems (CMS).
Image enhancement
Image enhancement is required only for images that are of low resolution or poor quality, which can impact optical character recognition (OCR) results. Image enhancement improves the capability to convert an image into digital characters. Our image enhancement capabilities include binarization, de-skew, de-speckle, noise-reduction, auto-brightness, and contrast.

Use the Pre-processor package actions for image enhancement. See IQ Bot Pre-processor package.

Document classification
Document Automation uses NLP, unsupervised and supervised learning, and OCR engines to classify documents based on their type and content. This process, when the document type is not determined in advance, enables efficient routing of documents to the appropriate processing workflows. Use the Document classifier or Advanced Classifier package actions to classify documents. See Document Classifier package and Advanced Classifier package.
Document extraction
AI algorithms are used to extract relevant data from classified documents. See Create a learning instance in Document Automation and Document Extraction package.
Document validation
Document validation is done by automatically applying regular expressions checks (regex), rules, and scripts to assess, match, and manage the extracted data for accuracy and relevance. See Validation rules in Document Automation.
Human-in-the-loop validation
Automation Co-Pilot capabilities enable exceptions to field and table value identification to be addressed directly in the application. The Automation Co-Pilot validator provides a user-friendly interface to highlight the errors or warnings in documents. The validator displays a red outline for fields that need validation. Users can validate the data for such fields and submit the documents for reprocessing. When all the validation errors or warnings are fixed, the document is successfully validated and sent for further processing.

For specific document types, validation uses supervised learning to provide a rapid feedback loop and fine-tune AI training by correcting data via human input.

Automation Co-Pilot can be integrated with third-party applications such as Microsoft Teams, Salesforce, and ServiceNow for users to perform human-in-the-loop validation and validation tasks within such applications.

Approval task
Automation Co-Pilot provides the capability to manage and assign approval tasks to users.
System of record update
The final step involves sending the extracted data into downstream systems. This data can be used to trigger business processes, update transactions, or complete cases. Usually, the data is pushed to enterprise systems such as document management systems (DMS), enterprise resource planning (ERP) platforms, or other systems of record.