Trapit, a Palo Alto startup that has launched a popular app, has an origin story that sounds like something Robert Ludlum would have dreamed up as a parallel plot for one of his Bourne spy novels. The company is built around patented intellectual property that was initially developed in a $200 million artificial intelligence project by the US Defense Advanced Research Projects Agency (DARPA) and R&D powerhouse SRI International. The project was created in the aftermath of September 11th to help source intelligence signals from the web, but it also turned out the artificial intelligence algorithms which came to be known as SIRI, acquired by Apple in 2010.
Trapit is intended to address a fundamental problem of the information rich world—”too much data, not enough time”—by giving readers the ability to craft custom “traps” to collect and portray content on literally any subject. Gary Griffith, chief executive of Trapit, spoke to Lean Back 2.0 from his Palo Alto office.
You were an aerospace engineer, a nuclear submarine officer and an IBM executive. How did you end up in a Palo Alto startup? When I was still at IBM back in the early 1990s, I got fascinated with this new phenomenon called the internet, so much so that I decided to leave IBM and start a company. Strangely, given my background, it was a gaming company called heat.net. I found the intersection of content and hardware technology fascinating. We built a company and had a few million people playing Quake together. While we’ve recently seen Facebook and Twitter emerge, social isn’t really a new thing. People forget about ICQ and bulletin boards and how social online gaming is. We sold heat.net to Sega. And since then, everything I’ve done has been about communities on the web.
What is Trapit? It’s a fascinating story. After 9/11, when the intelligence community was sifting through all of the data, they concluded that, if we were able to put that data together on a real-time basis, theoretically the attack could have been prevented. All the signals were there, but they were so buried in the clutter that humans could not pick them out. That’s when DARPA came to SRI to develop machine learning and artificial intelligence to process this data, and present it such that people can make decisions around it. Five years and $200 million later, they introduced CALO—Cognitive Assistant that Learns and Organizes. Out of that body of research, in addition to specific applications for various intelligence agencies, came SIRI. Two years after SIRI emerged, my co-founder Hank Nothhaft, Jr had been approached to talk about some online content technology that the project had developed. He found part of this CALO project that was essentially a news reader developed for the Navy. He was fascinated by the algorithms, and he called me in and we figured out what would need to be done to commercialize the basic algorithms. There were in fact a number of patents that SRI had filed in this general space. SIRI took some of the patents, Trapit took three patents, and both SIRI and Trapit have access to other patents from the project on a non-exclusive basis. Since then we have added another four or five patents that have made some drastic changes.
I use Trapit and I love the technology as a consumer. But it sounds like you have a vision for the company that goes beyond consumer reading apps. What you’re using is what we consider a reference application. What we do and what we think we’re really good at is having the world’s best personalized recommendation engine. That’s what we know and what our engineers do. We can sift through a bunch of information and deliver content that is personalized. We don’t start with the assumption that we are going to categorize the web; it’s all 100 percent personalized. For instance, I could read a story from The Economist. If I enjoy that, I can build an entire trap on that story. The discovery element of Trapit provides a pretty interesting way to consume content.
How are you working with publishers? We are beginning to create partnerships with publishers. We recently announced a partnership with Astro, a media conglomerate. We have a publisher partnership that is in place that will be announced in November. This publisher will take the Trapit application and it will be branded by that publisher. They will take at least one of their properties and set up a couple of standard traps, streaming their content into that application. They will blend that with their own content. This is one of the things that the content management tool will allow.
We started talking to publishers a few months ago with the idea that we could help them with their business models by helping them with a solution for their own content. It would enable them to sell individual stories. As we got into that, we saw how much these publishers thought the tools were very interesting, which we took for granted. Now we approach publishers with these CMS tools. We aren’t actually selling them yet—we’re in beta. But generally publishers have a mountain of content.
For instance, an Economist reader may not know that, if he is on one blog, there may be some very relevant content in other parts of The Economist. These tools would help by exposing other sources of Economist content to the reader. They could also be helpful in doing the kinds of things that we do—like posting relevant information to relevant users on Twitter or creating dynamic topic pages. For instance, we have the fiscal cliff issue that is looming. The Economist may want to have a dedicated topic page or an application about the fiscal cliff that is pulling in information that is relevant to this issue from across your sites. A lot of this can be done today, but it is very labor intensive. Trapit will allow much of this to be done automatically.
You’ve mentioned that your business model is focused on the enterprise. How will Trapit help businesses? When I look at the consumer space, there is a huge amount of competition in this field, but when you look at enterprise, no one really seems to be attacking the issue of information overload well in corporate America, particularly when it comes to blending public information with private information behind a firewall. Who is the leader in enterprise search or enterprise discovery? There aren’t a bunch of household names doing that. That’s where we have an unfair advantage; regardless of the topic, people can be as granular or as broad as they like—that’s the advantage. Right now consumers are cutting the sources millions of ways. We can do that for businesses.