A Page Approach to Capture

by Lowell Rapaport

If all you have is a hammer, then everything looks like a nail. Similarly, to many document and data capture systems, everything looks like a batch. In some cases, this mindset forces you to presort documents before you start scanning, and getting back to a document or transaction perspective can get complicated.

InputAccel from ActionPoint, San Jose, CA, has long offered a page-level approach to document imaging. In place of one-way processes and single-destination exports, the software lets you split and merge tasks at a granular level. You can also interact with line-of-business, ERP and CRM systems as well as back-end repositories.

The cornerstone of InputAccel lies in its open, flexible "capture flow," and version 4.0, introduced in December, adds new features that capitalize on this open, extensible architecture. This client/server software relies on clients to handle human interactions, such as scanning, indexing and quality control, while servers control the workflow and unattended operations such as image processing, recognition and data and image export. The clients and servers can coexist on the same machine, but in most InputAccel systems, modules are distributed over several machines across an enterprise, improving scalability and flexibility. Scanning and validation, for example, can be handled at any location on a LAN, WAN or virtual private network while still being coordinated with the workflow on a central InputAccel server.

InputAccel's capture flow routes work in progress to different users and even separate systems for processing, exception handling and supervisory approval. In many cases, the system brings steps normally applied at the back end of a document or content management system into the front-end scan and capture process.

For example, the system can deliver data to ERP, CRM and accounts payable systems as well as content repositories and databases. This approach can circumvent data entry steps that would otherwise take place by inefficiently copying or keying data within enterprise applications. Even more powerfully, interactions with enterprise systems can add intelligence to content before images and data are committed to the back end. The latter scenario was attractive to Imaging Acceptance Corp. (www.imagingacceptance.com), a Warrenton, VA-based service bureau that scans more than 2.5 million pages per day.

"[InputAccel's] workflow routing can be controlled by dynamic data coming from outside the system, such as one of our customers' databases accessed over the Internet," says Thomas Ketcham, Image Acceptance Corp.'s chief technology officer.

Ketcham cites one government agency that took advantage of this functionality. "[The customer] gave us a project that included classified documents," he explains. "InputAccel was able to check an external database in real time over the Web in order to identify the classified data. The capture flow was then able to reroute the classified images to more secure systems."

InputAccel supports an eight-level hierarchy that goes down to the page level. Individual folders, documents or pages can be split out of the batch and independently routed. Content remains in the context of a tree structure while also gaining throughput advantages by sidestepping unnecessary processes and eliminating the delays inherent in waiting for complete batches.

InputAccel offers two ways to create capture flows. A wizard is available for building simpler workflows while complex, customized workflows can be created in an integrated VBA (Visual Basic for Applications) development environment. Capture flows are also extensible through the use of custom add-on modules. The module architecture is a published standard, and ActionPoint provides a software development kit with sample code, a version of the PixTools imaging toolkit from Pixel Translations (an ActionPoint subsidiary) and the InputAccel API.

Safeco is among the customers attracted by the extensible capture flow. A national insurer based in Seattle, Safeco scans some 40,000 pages per day — about 80 percent of that volume is made up of claim forms.

"The customizable workflow lets us use different levels of validation on our forms," says Steve Minor, a Safeco systems analyst. "Some workflows send images to primary validation only, others to secondary validation and some to both. With VBA, we can develop workflows the same way we develop other internal applications. We were also able to use InputAccel's SDK to customize the validation engines and the integration modules we use to bring [images] into our content repositories," Minor explains.

Among the features introduced in InputAccel 4.0 is improved support for data export through a new XML exporter module. The module lets you deliver XML instance documents that you can set up by mapping to a schema or by inheriting a schema from an existing XML instance document. This approach speeds the process and lowers the cost of delivering XML-formatted data to multiple enterprise systems.

Other new features in 4.0 include improved integration with Adobe's Acrobat Capture PDF image conversion software and an auto annotation tool that automates image mark up. One application for auto annotate is to redact confidential information on an image at scan time, which can be used to protect customer identities, medical data or credit card numbers. Lastly, InputAccel is among the first products to support the latest high-speed color scanners capable of delivering color and bitonal images simultaneously.

What's surprisingly not available in InputAccel version 4.0 is any sort of thin-client or even plug-in client options for distributed scanning, validation or administration. While competitors including Kofax (www.kofax.com), Captovation (www.captovation.com), Captiva (www.captivasoftware.com), ReadSoft (www.readsoft.com) and others have all introduced thin- or plug-in clients for one or more of these functions, ActionPoint maintains that existing technologies and bandwidth constraints make it impossible to support production-level processing without client software.

Nonetheless, the company acknowledges that it is currently working with Prevalent (www.prevasoft.com) of Colorado Springs, CO, to integrate that vendor's QuillixWeb browser-based scanning technology.

Among the high-volume customers that populate the InputAccel user base, what matters most, says ActionPoint, is the flexibility, scalability and customizability of its capture flow. While the 4.0 upgrades are evolutionary rather than revolutionary, the core platform and open, customizable capture flow are appealing for high-volume, enterprise-level deployments.