Universal Mind Helps Viziant Develop Its Knowledge Mining Product
"We needed to have it actually working and demonstrable in a technology that's still not widely available. We absolutely could not have done it without Universal Mind."
High-tech startups face a critical milestone if they hope to attract funding. The "Wow!" factor is important. But what really impresses investors and customers is proving the product actually works when initial buyers are ready. Selling those first customers opens up important opportunities for further revenue and credibility - two vital ingredients for early stage growth.
That certainly was the case for Viziant, a private equity-backed startup bringing to market a new knowledge mining and visualization product for government and commercial enterprises. A technical proof-of-concept scheduled for the end of April 2007 would be key to Viziant successfully landing its first customer - a U.S. Department of Defense agency.
The only problem: by mid March the software's user interface - the part of the product that prospective customers actually see - needed to be developed to spec under tight deadline. That part of the rapid development project was especially critical since one of the product's most appealing and complex features is its rich visualization. So Viziant called in Universal Mind.
Viziant Senior Vice President, Jeannine Bartlett, recalls their first meeting. "It was the week of March 12th. Proof-of-concept happened April 30th. We asked them to do all of the UI development for us. Since this was our very first product rev, it was make or break time."
"We needed to show polished results."
Universal Mind immediately dedicated a team of five. "Those guys were able to jump pretty quickly into the middle of what we needed developed," says Bartlett. "We had potential opportunities with U.S. government agencies. We needed to show polished results within a specific timeframe just to be considered for some of these projects."
Why Universal Mind? "We were looking for the types of technologies that would be appropriate yet lightweight and look very fresh and new," Bartlett says. "That led us to Adobe. Universal Mind looked like really the only company out there with the depth and breadth of expertise to be able to execute that."
Viziant's product allows multi-lingual users (English and Arabic, for example) to rapidly identify entities in a database matching concepts specified as people, places, things, and events. Results can be viewed hierarchically or in a "spring graph" diagram mapping relevance among concepts within a frame of reference, such as "New York." The tool would be extremely helpful to an analyst looking to quickly and intuitively map all data related to a particular entity from among possibly terabytes of raw information.
The desire to offer users a rich interface was one reason Viziant selected Flex and Universal Mind. Another was the need to deploy that interface easily on user desktops and to drive it with minimal overhead across the network.
Rich and Lightweight
"We looked at several different approaches," Bartlett says. "We looked at building this from the ground up in Java. That certainly would have given us a great deal of control over the UI but it would have taken many more man months than necessary with Adobe Flex. And Java tends to be a little bit fragile. We also looked at .Net environments and found that it was simply too heavy on the client side and was too restrictive in many cases. And it did not really have that leading edge web look and feel that Adobe does."
Because Flex UI clients are so lightweight, there is great flexibility in where customers can deploy, Bartlett says. "Flex also offers many ways to minimize traffic between servers and to maximize the number of clients running heavy visualization without grinding the system to a halt."
Finding and exploiting those opportunities, says Bartlett, was just part of Universal Mind's job. "Universal Mind was responsible for taking our entire UI design specification and implementing it in Adobe Flex; for tying those clients to the BEA WebLogic server over Adobe Flex Data Services; and for just overall participation as our experts on the team for that technology."
Avoiding data bottlenecks was key. "We have potentially hundreds of thousands of users," Bartlett says. "So we've had to pay particular attention to scalability issues, parallelizing issues, and also be very sensitive to how much data is going over the wire at any one time and how many times we're going to have to hit both the database servers and the applications servers."
But avoiding bottlenecks in the application development process was of equal importance.
"A lot of this depended on the speed with which we could bring this together and show it actually working in a technology that's still not widely available on the market with either full-time hires or other contract companies. Just based on their speed and agility we absolutely could not have done it without Universal Mind."
Universal Mind streamlined development in other ways too such as by using existing third-party components rather than write everything from scratch. "They were able to save quite a bit of time," Bartlett says, "simply based on their breadth of knowledge of the Adobe Flex community and what has already been done out there."
She cites a date charting component as one example. "They leveraged that pretty quickly. Even the spring graph was an open source component, which we'll probably have them customize heavily for the final release."
As for integrating Universal Mind into her team, Bartlett says things could not have gone more smoothly - even though most Universal Mind work was actually performed remotely.
"Our deadlines are their deadlines."
"We really look for people who can become part of our team - work side-by-side with us. And they've been absolutely stellar at that. Our deadlines are their deadlines. They bend over backwards every time. My developers work hand-in-glove with them on an hour-by-hour basis. And they've just been superb."
Superb - and also on time. Thanks to this joint effort, Viziant did in fact make its first technical proof-of-concept on April 30th, when it landed its first customer.