One on One with Francois Bourdoncle of Exalead

Francois Bourdoncle is the co-founder and Chief Strategy Officer for enterprise search vendor Exalead. He contributed to the development of AltaVista, the world's first large scale Internet search engine. He says of his AltaVista experience, "As you might expect, I've taken scalability seriously even since. In particular, I was exposed to 64-bit and multi-threaded computing very early on, and have always built systems on top of these technologies--Exalead's search technology is no exception." We had a chance to talk with him about what Exalead is doing and about enterprise search in general.

FCM: What changes have you seen in enterprise search in all your years in the business?

FB: Some people say that enterprise search is dead (or commoditized). This is probably due to the fact that search as a standalone application (e.g., Intranet search, or search on top of a single Content Management System) has proven to deliver little business value and that Google and Microsoft are quickly capturing this market anyway. 

However, the amount of data produced by large corporations is booming (four exabytes of unique information were produced last year--that is 4,000 petabytes, or 4 million terabytes, or 4 billion gigabytes), and scalability is quickly becoming the number one requirement of high-end search systems.

I actually believe that enterprise search is still in its infancy, at least when you think of "search" as an Internet-inspired infrastructure for designing business applications. Of course, most of the building blocks are already there--faceted navigation, linguistic and semantic technologies, grid-computing technologies, dynamic clustering and categorization, to name a few. What is very new, however, is the way to combine all these technologies into a coherent infrastructure that can be applied to all sorts of business contexts and at any scale.

When I think of it, enterprise search systems are becoming more and more similar to database systems, but with a twist: search systems are optimized for accessing information (not for making transactions), and for being used by humans (not machines). Instead of relying on hard-to-use forms, search-centric applications use a single query box, and this makes a major difference to end users.

FCM: How important is Web 2.0 in the enterprise and how does your new CloudView product help incorporate increasing amounts of data that are stored outside the enterprise firewall?

FB: Web 2.0 technologies are very important since they are at the core of the applications that people use every day on the Internet. Most websites provide a search box, and people expect that. But people also expect to access information in the enterprise with the same ease-of-use. Today, business applications are built very differently, typically on top of relational databases, and are unable to deliver the level of usability that enterprise users demand.

So Web 2.0 technologies, and search technologies in particular, are key to mass-adoption of business applications in the enterprise. Business intelligence is a very interesting example, since modern search technologies lend themselves very well to developing simple BI applications with user interfaces that are so easy to use that almost anyone in the enterprise can use it.

What is clear also is that business applications such as CRM now need to give access to a lot more information than they used to. For instance, how about indexing email sent by customers through the enterprise website, or to customer support? Or, for another example, extracting product names and causes of failures in customer emails can greatly help increase the effectiveness of a CRM system

Exalead's CloudView product range aims at extracting information wherever it is stored (databases, mail systems, even the web if necessary), structuring it so that applications can use it effectively, and making this semi-structured information easily searchable.

FCM: Many enterprise search users want a Google experience. Why do you think this is and do you see vendors, other than Google of course, trying to make that happen?

FB:I think that what people call the "Google experience" has more to do with the ease-of-use of Internet applications in general than Google specifically. 

What is true is that most people now want direct access to applications through search...and they want a result fast. But we all know that the magic sauce that made Google ranking so good on the Internet (link structure and contextualization of links) is not directly applicable in the enterprise. 

What is key in the enterprise is the ability to guide users in their search process and also to provide technologies to help users not miss a very important piece of information (e.g., using approximate matching). For instance, it is essential to have the ability to search an address book using approximate matching, since one often forgets the exact spelling of names. Also, structuring information by extracting facts and the relation between these facts is much more important in the enterprise than on the web, even if the underlying technologies (RDF, OWL, etc.) have been introduced to implement the semantic web.

FCM: What do you see as the biggest challenges for enterprise search in 2009 and how will you meet them?

FB: I think that 2009 is going to be a turning point for enterprise search. Search as a standalone application will become less and less important, and search vendors in this area will probably struggle to survive.

However, in the current economic environment, large corporations will look for new ways to reduce their operational costs, and will most probably delay large-scale (and expensive) IT projects, such as ERP or CRM migration projects. But at the same time, IT departments will still have to deliver new applications and functionalities. I believe that search-based business applications can help large corporations drastically reduce their IT costs, and that will be Exalead's challenge for 2009.

FCM: What impact do you think semantic search will have on enterprise search (and search in general) in the future?

FB: Semantic search technologies are essential for providing search systems that can "reason" about the information that they store. They are also essential for building search systems that provide answers as opposed to endless listings of search results. Exalead's high-end offering, called CloudView 360, is precisely that: A semantic repository (that we call a "semantic cube") layered on top of our full-text search engine.

More generally, semantic web technologies will prove just as useful on the web and in the enterprise, making the use of data gathered on the web in business applications a reality.

Related Articles:
One on One with Content Management's Movers and Shakers