Speaking to Techworld, CEO Peter Wallqvist explains how Ravn's AI robot, Ravn ACE, is changing the way companies approach data processing, saving them time and money.
Ravn's Applied Cognitive Engine, or Ravn 'ACE' can search, read, interpret and summarise vast amounts of unstructured data, 10 million times faster than its human counterparts.
Ravn Systems CEO Peter Wallqvist says: "We realised that when you do [data] searches you’re not always after just a big bunch of documents, you’re actually after the information that resides within those documents.
"So our research and development efforts were very much focused on being able to extract what is of interest to [customers]," he adds.
Founded in 2010, Ravn Systems specialised in next generation enterprise search, moving into artificial intelligence in March 2015. Impressively, Ravn has built a brand without any investment, growing entirely organically.
"Myself and the other co-founders used to work at a company called Autonomy (now HP Autonomy) and sort of felt back in 2009 that we could take the next step in technology and unstructured data processing and artificial intelligence and those sorts of things. So we all left that company and we started Ravn," says Wallqvist.
"What we realised when we started showing this to people was that the ability to actually understand things within documents was more important than even searching for information," he added.
How does Ravn ACE work?
Ravn ACE retrieves specific information by sifting through folders, files and documents, essentially producing meaningful information from unstructured and disorganised data. This process mimics what a human would do to extract important information from a document, but at a much faster rate and more accurately.
"Our core platform now supports this ability to understand what’s important, extracting that and putting it in a different context. These things are quite domain specific so if we’re dealing with, let’s say a real estate practice in a law firm then we extract certain types of things within that document set.
"Throughout the years, we’ve developed a more advanced degree of understanding and that’s when we started touching on the deep learning type of algorithms to actually understand and break down individual wording and inferences to understand the meaning of this content," says Wallqvist.
Ravn ACE is a relatively generic software model, lending itself to index any manner of documents over countless sectors. So within this there are domain-specific configurations.
"It’s very difficult to create a robot or any kind of AI platform that does everything in a very general domain. That’s the holy grail of AI, but of course, that doesn’t often work that well so we tend to have a model where we have a very easy way of teaching that robot to do a particular thing.
"For instance if we go in and do work on a financial transaction there’s no point looking for something that’s not going to be there. So we tend to use the simile that we have a very clever, but empty robot brain to start with but then we teach that to do certain tasks easily," says Wallqvist.
"Having done this for many different industries, we are starting more and more to deliver specific pre-built robots that are already very good at one particular task," he adds.
Where is Ravn ACE being used?
International law firm Berwin Leighton Paisner (BLP) has successfully implemented Ravn ACE to speed up mass data processing. This has resulted in a lighter workload for employees, more accurate data readings and an all round boost in efficiency.
"We realised that there were so many processes in law firms but also in other areas that deal with just reading documents and extracting interesting details," says Wallqvist.
"Before we had a selection of particular relationships with a bunch of firms and companies that we worked together with and our challenge now is to scale that so we can support many, many different clients and desires.
"So that’s what we’re busy doing now. And at the same time looking at other industries and verticals where our technology is appropriate," says Wallqvist.