AI Behemoth IBM Enters the Cognitive Capture Market
Document capture is nothing new at IBM. The acquisition of FileNet Corp in 2006 and Datacap in 2010 provided, at the time, noteworthy additions to IBM’s voluminous portfolio of offerings. But the document capture market has changed dramatically since 2010 and is set to evolve much further over the next few years.
Indeed, the traditional world of Document Capture is evolving into what we like to call “Cognitive Capture”; the key term here is the word, Cognitive. The addition of Artificial Intelligence (AI) radically transforms the ability to automate the process of extracting structured, actionable business information from documents. Therefore, it is not surprising that IBM, a company already synonymous with AI, should exploit those technologies and skills for document processing.
Following the IBM Automation Document Processing launch a couple of weeks back, we managed to get an early look at the product and plan to publish a Vendor Vignette shortly. For now, I thought I would share a quick take on what we saw and heard.
The ability of Cognitive Capture to liberate structured business information from documents delivers tangible value when combined with other technologies, e.g., BPM, RPA, etc., many of which already reside within IBM’s stable. At launch, IBM Automation Document Processing offers integration with FileNet Content Manager and IBM Enterprise Records Manager. These are both content repositories that can benefit from the increased visibility of the information contained within the documents processed and subsequently stored on them. There is also an integration to extend the capabilities of IBM’s Datacap capture offering. As befitting a component of IBM’s Automation Cloud Pak, efforts to integrate with IBM’s Business Automation Workflow are ongoing (it already utilizes the same low-code Application Designer UI.)
We expect that other integrations will follow, incorporating, for example, IBM’s newly acquired Robotic Process Automation offering. Of course, all such data-hungry platforms thrive on structured business data; consequently, the ready availability of such information will significantly enhance the level of automation.
IBM plans to include, at general availability, several pre-trained AI-based algorithms enabling the extraction of data from a variety of documents, including:
- Purchase orders
- Bills of Lading
- Utility Bills
- Selected Tax forms
Customers can then further enhance these algorithms by adding ancillary data fields and conducting supplementary training with sample documents, and train the system to recognize additional document types. Target uses cases for the product include:
- Insurance – Enrollment, underwriting, claims, and policy servicing.
- Financial Services – Commercial & consumer lending, new account opening, KYC, AML, dispute resolution, and account service requests & changes.
- Public Sector – social services eligibility and financial verification.
- Accounting & Finance – invoice processing and order to cash.
Overall, we see this as a positive move to enhance both IBM’s content and automation capabilities. The decision to include several pre-trained algorithms is noteworthy but hardly surprising given the breadth of IBM’s vertical domain expertise and its customer base. This launch represents a pretty significant upgrade to IBM’s capture portfolio, and we expect that it will generate interest from customers partners and, of course, competitors in the market. Again, we will be publishing a more in-depth Vendor Vignette soon.