Benoît l’Archevêque: I’ll give you a very quick example. If I do a search on Italian, red, and car, you see a Ferrari. What we have created is a new dynamic knowledge graph where we only store words once. If I have 500,000 bottles of wine, I’m not going to store the word wine 500,000 times because I have 500,000 different bottles of wine. I will store wine once. I will store red once. I will store the words that are not common to create this other group of products. We were able to build a new way of structuring data. We can now take homogeneous and heterogeneous data and deconstruct and reconstruct in real time. For us, it’s a matter of seconds.
After that, the second problem that I wanted to solve was to be able to surf on the mobile. Right now, what people do is take the web and bring it to the mobile – the same way advertising was brought to the web without thinking about it. We had billboards. They put billboards on the web thinking that was the thing to do. When you look at a website, it is a big search engine. When you actually divide your information, it’s actually just a way of structuring the data so people can find it themselves. Even if there are a lot of dead ends, it is still a big search engine.
Sramana Mitra: We are talking about context-specific search basically. Every search has a different context. You will see that there’s a piece that I wrote in 2007 called Web 3.0 = (4C + P + VS). The four C’s are context, community, commerce, and content. P is personalization and VS is vertical search. In that model, we’ve seen a lot of companies come up in travel, real estate, and commerce. What you’re alluding to is that this kind of vertical search capability is not available dynamically. Help me understand what it is that you’re doing with Azzimov.
Benoît l’Archevêque: Everything you said about your article, I totally agree with that. Contextual search is important. If you’re in one vertical, it’s pretty easy to create something that looks rather intelligent. But there’s no intelligence in it – like Siri. Even if it looks intelligent, it’s only raw processing power of a lot of possibilities. What we actually try is to mix different types of data. If you want to do a Google-type product portal, you need to be able to mix all the different structures.
Sramana Mitra: You are not doing vertical search but you’re doing horizontal product search?
Benoît l’Archevêque: I’m doing both at the same time. Imagine more like a sphere. Imagine that the angle in which you’re going to enter the data is going to be from your view. The system will adapt to that. If you’re looking for red, I’m going to ask you, “Are you looking for a wine or furniture?” I’m not going to get you into one vertical. You can go in any direction at any moment whenever you want.
Sramana Mitra: What is the go-to market strategy for what you’re doing? What are you selling? Are you a portal or are you selling software?
Benoît l’Archevêque: The company is actually both. The business model was to create a Google or portal-type where you can find any type of product that you want in the world. People can go search what they’re looking for because it’s product-centric. If for example you’re looking for a table, I can see all the merchants that are geo-localized around you that offers that table or a similar one. My business model is there. As a consumer, you ask for a quote. This quote is sent to maybe 10 retailers that are around you. I provide those retailers a name, address, email, and an intent purchase in real time. Retailers can actually answer and buy those coordinates. What I’m trying to do is recreate the brick-and-mortar environment where we can negotiate and talk with someone online. Everything was done to fit the mobile.