For years, document capture technology has been fairly limited in its application. Initially, it was used for tax, insurance, and other types of “structured” forms that had the same layout every time. Then, in the early aughts, we came up with more flexible technology that basically worked well on variably structured forms (same information in different places), with invoices being the most prevalent example. Automated classification of documents in mortgage packages has also developed into a healthy market.

 

But, the biggest challenge with all types of document capture has always been the set up. In the early days, it took template drawing and testing and redrawing and more testing. Of course, you can throw in techniques like regular expressions, scripting, database look-ups, etc. And then more testing. It’s not uncommon for capture applications to cost in the tens, if not hundreds, of thousands of dollars and for the set up to take months. And, then if you want to introduce a new form….

 

In recent years, as machine learning has been introduced into document capture, one of the promises has been reduced set-up times. Leveraging machine learning, a user should be able to feed an application a few samples, make some corrections, and viola, they should be up and running. The “learning” capabilities should enable the application to automatically improve over time, typically based on corrections entered during the QA process. New documents should be able to be onboarded through simple QA interaction.

 

Of course, as we said, this has been the promise. The reality has often proven to be that manual assistance in the way of scripting and regular expressions was often still needed to optimize performance, and there has always been the danger of giving the learning mechanism “too much information” and confusing it.

 

Recently, however, we’ve been hearing stories that lead us to believe that maybe machine-learning driven capture is ready for prime time. The latest news comes out of Grant Thornton LLP, an audit, tax, and advisory firm, that is listed as the sixth largest accounting firm in the U.S. Grant Thornton recently announced that is has formed a partnership with Ephesoft to introduce its machine-learning driven capture as both a product and a service.

“We are leveraging Ephesoft in our tax, as well as our innovation practices to enable solutions for our customers,” said Nick Vellani, advisory practice leader for the central region at Grant Thornton. “The fact that we can get an Ephesoft application up and running in two to four weeks—it enables us to do truly agile development and has been a game changer when compared to capture products we’ve worked with in the past.”

 

Within its tax practice, Grant Thornton has implemented Ephesoft to capture data from K-1 forms, which are issued annually to investors in partnerships. “We have previously been capturing data from K-1s manually; with Ephesoft, we are now able to capture it with a very high rate of automated throughout,” said Vellani.

 

In its innovation practice, Grant Thornton is implementing Ephesoft in a variety of solutions. “There is value in extracting information off documents in the initial stage of almost any sort of automation process,” Vellani said. “If you look at RPA, or AI, or digital transformation—you have to solve the initial problem first. We found that Ephesoft does a great job solving the data capture and extraction problems that a lot of our clients are facing.  For example, we use Ephesoft to extract invoice information in A/P processes. We started with header/footer and then had a client ask us if we could do line item details. We are able to extract data from single-spaced invoices that are 20 pages long.”

 

Vellani added that there have been two significant changes in GAAP, related to ASC 606 and ASC 842 forms used to report revenue recognition from leases. “These changes really require you to dig into the leases and agreements in ways like never before,” he said. “We are using Ephesoft to exact information from these forms and put it into a spreadsheet or data share that enables analysis for proper revenue recognition.”

 

Grant Thornton has the option of deploying Ephesoft on premises or setting up calls for its customers to the Ephesoft cloud. Grant Thornton is also considering hosting a version of Ephesoft on its own cloud. Vellani said that flexibility and footprint are the two biggest differentiators of Ephesoft compared to other capture products.

 

“We have flexibility as to what environment we want to run it in, and we can leverage the tools to address a wider variety of documents,” he said. “Those are game changers for us. It’s a true enterprise solution. We can easily develop a solution to solve our customers’ business issues. As opposed to six months or a year-long process, like you have with other capture software, it takes about a month to get it done.”

 

Vellani explained that Grant Thornton looks at capture as a piece of larger solutions it is implementing for its clients. “We start with the problem that the client is facing,” he said. “Let’s say we have a client receiving 450,000 contracts that they have to look at for revenue recognition purposes,” he said. “Ephesoft becomes a component of an overall digital transformation strategy designed to help them better monitor those contracts. We are looking to implement technology as a way to accelerate processes that need to be completed—either internally by our customers or through our consulting practice.”

 

Stephen Boals, SVP, business development and global alliances for Ephesoft, noted that it’s these types of deployments that help move capture up the value chain. “It helps move it from a commodity to the highest level of business operations,” he said. “Grant Thornton has great relationships with its customers, they know their problems and what solutions they can implement to address them. They are now providing solutions at the top of the value stack that incorporate Ephesoft.”

 

Added Brandyn Campbell, Ephesoft’s director of business development, global consultancies and system integrators, “Nick’s team are industry experts; specific to their clients, they have a high level of understanding of business processes. They understand the documents, workflows, and manual steps that are undertaken by these businesses, and they can take our software and leverage some of their own IP to create a solution that gives their customers a competitive edge.”