Sramana Mitra: So you are taking the fire hose of Twitter data for Datasift and contextualizing it, and then your clients do whatever they do on top of that.
Seth Redmore: Yes. What you do as an enterprise or entrepreneur is get a feed of data sets that is precisely tailored to your requirements and that lets you say, “This is just the stuff that I want to see.” That turns out to be a very interesting problem. It is a filtering problem. How do you filter all this data?
SM: Let’s do one more use case.
SR: Oracle acquired Endeca last year. Endeca is a business intelligence platform, a search engine–based business and BI platform. The platform is very flexible in terms of how it describes things and how information flows through it in a way that you can pull it out of it. They do a lot of stuff around insurance fraud, supply chain analysis, or manufacturing root cause analysis. We are the text analysis engine for that. If somebody has free text they are dealing with, and they are dealing with an Endeca/Oracle installation, they buy our software. So you can roll up questions like, What are the important phrases that people are using in the context of Twitter? There are two parts to it. One is the discovery: “Show me what is happening now.” The second part is, “I am going tracking.” Or, “We have these three different warranty areas for Twitter. They look pretty steady. Has any of them ticked up or ticked down over time?” It is about helping clients to understand all the mechanics notes that are coming back in or are in the manufacturing defector ports, so they can do the root cause analysis.
SM: Let me ask you how you work with Datasift. If a small startup company – I am asking you this because our audience is full of startup companies, and we see a lot of social analytics types of businesses – wants to start building a product using your technology as the enabling technology on the analytics side, how do you work with them?
SR: We have a number of ways that we can do it. The engine itself is off our API. It is an enterprise [type product]. You take it, you install it, and you code to it. We support a number of the major programing languages. From a business perspective, the big thing with small startups is cash and cash flow. We set up deals where we defer licensing payment. You sign your deal, and you don’t have to pay us for a year – you got it going. We do engineering over licenses for people. If there is no cash and they want to do that, we certainly entertained things like stock before. We realize that there is risk for both parties. For us, it is really a matter of “show us that you are serious, show us that you are doing something with it, show us that you are not just kicking tires.” If you can do that, we tend to be very accommodating with early stage companies.
SM: That is really good to know. Off the top of my head, I can think of one of our portfolio companies that is a very serious company, and I know that they do analytics applications in your domain. This could be interesting for them. I don’t know which text engine they are using, but I will keep this in mind.
SR: That would be great.
SM: The domain you are in right now is driven by social media analytics, and it is a very popular field of application. There are lots of different types of applications happening in this general area. That has caught my attention.
SR: The thing that we are really trying to work on is expanding globally. We have a number of languages other than English: Portuguese, Spanish, French, and German, and we are just getting ready to introduce Mandarin. We are looking for people who have their feet on the ground in those countries to either work for us, integrate us into a service, or who happen to have a share point practice or an enterprise search practice. Text analysis fits in really well if you have some sort of business practice around predictive analytics. That is what we are looking for in these smaller companies.