In part one, we took a look at how RPA marked a revolution in empowering businesses to solve problems associated with manual, data-centric tasks. But what about the documents? RPA is historically ineffective in automating document processing.
Enter CDA. CDA does the “head work” of understanding what the document or email is about, what information it contains and what to do with it. RPA and CDA work in tandem to automate the data-centric tasks and document processing that are present in many business processes. CDA encompasses three process stages that streamline business operations: Acquire, Understand and Integrate. For part two of our series, we’ll take a closer look at these stages.
AcquireRPA+CDA systems can both capture documents (CDA), and access electronic data (RPA). For CDA, the documents could be in a variety of formats—email, fax, folder, PDF and Office files, website uploads, MFPs, scanners and, especially, mobile devices.
Flexibility is key: your customers are unlikely to have the patience to adjust to different instructions, re-submit information or, worst of all, start over; therefore, your document submission process should be “smart” enough to allow them to switch back and forth between channels during the same process.
Embedded document capture capabilities are also important for mobile apps and capture-enabled websites, and should enable real-time capture and data display and allow users to correct any errors in the data before submitting.
The document has been acquired. What next?
This is where CDA answers the following questions:
- What is this document or email about?
- What information does it contain?
- What should be done with the document and the information?
Cognitive document automation uses a variety of artificial intelligence (AI) capabilities, such as natural language processing (NLP) and machine learning, to cluster, classify, separate, OCR, extract, and understand (human language) any type of document. Machine learning is a key component of CDA, easing the configuration and maintenance of CDA systems. Just provide a few samples of each document type, and CDA automatically knows how to classify and extract data from them―there’s no need to write rules or create rigid layout-based templates for each document type. If documents change over time, machine learning gracefully adjusts to those changes without human labor. Absent machine learning, traditional document capture systems become obsolete soon after day 1 of production operations, as documents inevitably change, requiring endless manual configuration effort to keep up.
CDA also employs natural language processing (NLP) to understand text-rich documents like contracts, correspondence, mortgage documents, and M&A documents―documents traditional capture systems couldn’t handle. Instead of employing people to perform low-value tasks like searching through paragraphs of text for key information across thousands of documents, NLP can be used to quickly and automatically extract key data such as contract dates, amounts, parties or addresses―anything of interest.
Simply put, CDA uses AI to automatically understand and learn documents, so people don’t have to.
Typically, the steps required for CDA to understand a document are:
- Machine learning
- Image perfection
- Classify documents
- Separate documents
- Extract information
- Evaluate results
- Manage exceptions