A SOLID FOUNDATION

Editorial Type: Interview Date: 10-2020 Views: 901 Tags: Document, Software, Capture, Cloud, OCR, Hyland, Kofax, Lexmark, Perceptive, Microsoft, Brainware, ReadSoft PDF Version:
Hyland recently announced a significant update to its long-established Brainware Intelligent Capture offering, including enhanced handwriting processing, security improvements and cloud integrations. DM Magazine editor David Tyler found out more from David Luzier, Hyland's VP of R&D and Content Services Engineering

David Tyler: The Brainware name has been around for a very long time: can you give us a brief update on where the technology has come from, and what your involvement has been in its development?
David Luzier: I've been in the software industry for over 25 years now, with a focus on capture technologies. In essence, I've been with the same company since 1995, though it may not look like it from my CV, as there have been a lot of acquisitions along the way! Brainware itself emerged in the late 90s, and even then I recognised the potential it had in terms of document classification and extraction - even then the solution wasn't template-based, which gave it an advantage over rival offerings when working with semi-structured and unstructured content. In 2006 we decided to spin-off Brainware from its then owners (SER) and set it up as a standalone company. At that time we had fewer than 40 staff - I found myself in charge of development, product management, marketing and also IT for the whole company; it was an interesting time, to say the least!

Six years later, with me as CTO, we sold the Brainware company to Lexmark/Perceptive. There were a number of changes over time, as you probably remember: the Perceptive (i.e. software) side of the business acquired several other companies. I was then responsible for all of the data capture products, and for developing a vision of how to bring all these technologies together. There was a lot of overlap of course, but each of these products had their strengths and weaknesses.

Most recently, in 2017 there was a split of the Perceptive software division from Lexmark, and that was how the Brainware/Preceptive technologies ended up as part of Hyland. I went with the legacy Perceptive business and have taken on a lot more responsibility since joining Hyland: I run the R&D department for everything outside of healthcare, so that's not just capture technologies, but also the OnBase platform, workflow, and of course a lot of other products that Hyland has acquired along the way as well.


"Whether handprint, cursive, freeform, machine print - we've even taken handwritten letters and the system is able to read it all and identify a claim number, a social security number, everything. The way that we've integrated Microsoft Vision allows you to essentially just point to wherever the service is - so again, fitting with our future vision for micro-services - and the system will call it there, whether it is in the cloud or on-premise. So there are real improvements from an OCR/ICR standpoint, and then where our technology comes in is with the extraction, internal and external validation. That is what really drives the business value."

DT: August saw the release of what Hyand is calling Brainware Foundation; how has the product - or the technologies behind it - changed as a result of all these changes/acquisitions over the years?
DL: It's an interesting question: ultimately it is the same base product - a lot of the technology around Brainware is in its underlying engines and algorithms for document classification and extraction, which have been designed to minimise human intervention. What we've done all along the way is to build continually on those engines and algorithms: finding ways to automate things like complex table extraction for transcripts, for instance.

Hyland has both natively developed and acquired different capture technologies over the years. Our long-term vision is to take the best of these capabilities and offer them as SaaS-based microservices: you pick and choose the components you need to build the solution that suits your business requirements - and we'll also we'll be able to offer out-of-the-box solutions for applications like invoices, transcripts, etc.

That same vision involves 'decoupling' these sometimes monolithic solutions and bringing the best of them together. Our strategy, though, is not to try to do this in one big-bang approach. For example, in this release we have a new algorithm we call Automated Learning Engine - it's essentially intelligent enough to 'watch' the user and observe when they make corrections and learn from that behind the scenes.

This takes us even further from the definition/template-based approach still used by so many vendors; what we're using here is something called a Dynamic Variance Network - even though information shifts and moves around, we can deal with it dynamically. This is one of the new engines we've developed as a microservice with an eye on the future SaaS offering I mentioned, but we've already implanted it into our lead capture technology in Brainware. So this is what I mean when I say that it's still Brainware from its original foundation, but we're adding a bunch of new capabilities on top of that, these new engines and algorithms. We're investing heavily in those areas that will help us to drive automation.

DT: I was interested to see that the new Foundation version boasts improved handwriting processing. This has been an area where capture/recognition technology vendors have over-promised and under-delivered for a very long time, I'm sure you'd agree. What is Brainware doing differently?
DL: This technology has really come a long way. We are talking these days about processing free-form handwriting. Interestingly, this is not something we've developed within Hyland: we're leveraging Microsoft Vision. You'll remember the old OCR engines, which worked better if written text was constrained (that is to say, written inside a box), or you would tell it where to locate the information and it would read that specific 'zone' - these solutions go way beyond that by leveraging the cloud. Microsoft of course has so much data at its fingertips to learn on - and so much compute power - that it is able to continuously improve its engines: it is genuinely very impressive.

Whether handprint, cursive, freeform, machine print - we've even taken handwritten letters and the system is able to read it all and identify a claim number, a social security number, everything. The way that we've integrated Microsoft Vision allows you to essentially just point to wherever the service is - so again, fitting with our future vision for micro-services - and the system will call it there, whether it is in the cloud or on-premise. So there are real improvements from an OCR/ICR standpoint, and then where our technology comes in is with the extraction, internal and external validation. That is what really drives the business value. We're really excited about where Brainware is headed, and feel that it will open up opportunities for a lot of organisations.

More info: www.hyland.com/brainware