David Tyler: Let me begin by congratulating you again on your recent win at the DM Awards in the Data Capture/Recognition Product of the Year category. What does a win like this mean to you and to your partners?
Manfred Traeger: We were very proud to be recognised by the market with this award: we've been working on our products and trying to improve them for over 20 years now, so an award like this serves to motivate us all.

Our strategy has typically not been to go for the 'big bang' with our product developments, but rather step by step. As an indirect business it is important for our partners to see that continuity and consistency. We have to be aware of the balance between innovation and partner enablement.

Within our solution business we are 100% committed to the indirect business model, working not just with VARs but also BPOs and Shared Service Centres. In effect there is nothing more important than our partners - and winning a DM Award helps to reassure the channel that they are working with a recognised technology partner, a truly competitive software vendor in the market.

DT: How do you see the recognition and capture market itself changing - and how does that affect your product development plans?
MT: It is an important concept for us that 'everything starts in the mailroom', whether that is literal or symbolic. It is well established now, of course, that documents can mean much more than just physical paper documents. The definition has extended to include all kinds of electronic content from emails through voice and even video. This comes back to my earlier point about how we focus on gradual step-by-step enhancements to our products.

We started with the documents that are easiest to process - forms. Then we moved onto semi-structured documents and then unstructured documents, represented in application terms by areas like Accounts Payable. The next step is Order Processing: orders and delivery notes are much more complicated than invoices to automate, partly because the business rules behind that process tend not to be as 'strong' as those involved in invoicing. So we've done a lot of work behind the scenes to optimise the algorithms, to improve how we reliably extract that information.

While Accounts Payable is a kind of 'bread and butter' business for many in the industry, we are already moving towards more Case Management environments. These days we need to focus on all the business documents coming into an organisation - documents that are more unstructured, not sorted or classified, related to more diverse business processes: this is how we see the market shifting.

DT: From a technology perspective, does this progression in the market mean that you have to be more aware of the details of different business processes?
MT: If you look at a digital mailroom that is (in insurance, for instance) feeding 50 or even 100 different business processes, of course every business process owner would like to get all of that information automatically extracted. It would be far too complex - and expensive - to configure traditional capture systems to handle all of these different processes. The approach - and the technology - then needs to become more intelligent.

We are addressing this in one respect with a 'hybrid forms' approach. Typically we used to differentiate between structured, semi-structured and unstructured forms - but that is only because of how we worked in the past, and how the recognition software market has developed around the challenges of the time. But perhaps it no longer makes sense to differentiate in that way? You can use forms-oriented data extraction technologies, but you might also find it useful to use freeform extraction technologies as well - our solutions need to intelligently decide what technologies best suit each task.

It comes back to consistency - from a user perspective, they expect one homogeneous API, one validation system, one user environment - they don't want to be switching between different systems for different types of document. It's about using the features that are most helpful in terms of getting the data you need extracted, quickly and accurately.
More info: irisdatacapture.com