An Interview with Olivier Lefassy
I took a hard look at PolySpot at the Intrernatioal Online Meeting in London, England, in December 2007. On a recent trip to Paris, I caught up with the managing director of PolySpot, a former investment professional and business executive, Olivier Lefassy. Mr. Lefassy and I spoke at La Tradition. This eatery is different from the hot dog joint in Harrod's Creek, but everyone was nice to me even though I ordered fizzy water, not wine.
The PolySpot system provides information access functions. Key word search is available, but the focus of the company is on information access. One distinguishing feature of the system was its intuitive interface. Another was the company's rapid growth rate, over 100 percent in the last 12 months.
The full text of the interview appears below:
I am surprised by the number of business intelligence, search and content processing companies in France. For example, there's Sinequa, AMI (formerly Albert), Exalead, Digimind, and Lingway, among others. How can all of you possibly survive in a competitive market?
Yes, I think that France has many excellent technologists. In the 6th century, these types of thinkers would have worked in other high-tech promising businesses. Today, search is such a big problem for so many companies, there's enormous opportunity. Just like people in the wine business.
Yes, my analogy is like the firms working to create what we call Grands Crus. So why will all these wine makers survive? The reason is demand. And in France we have the right ingredients to deliver great products with world-class quality. PolySpot is like one of these Grands Crus. We have the next generation platform from which to build from with wonderful expertise in our team.
Olivier, I never thought of search and wine as similar. Are you pulling my leg because I am American?
Yes, yes, of course. But I am also very serious. Remember, some wines are excellent with cheese some other good with fish…The same with information technologies.
Sometimes customers needs will find satisfaction in a mix of several technologies, each one bringing its own level of expertise. For instance we do not compete with Digimind, on the contrary we have started to use the synergy of both solutions to provide targeted prospects with a high level of satisfaction as far as Competitive Intelligence and Search are concerned.
Where did the PolySpot technology originate?
In the early days, At PolySpot we tried to start thinking “smart and small”. That means that we tried to put efforts on key differentiators such as the management of the user interface, our APIs, and our full Web administrative console.
On the indexing side we chose to integrate Lucene because it is was and it is still an enormous and very active community. Plus, we think it does a good job of basic indexing.
What company can afford to have many engineers people working on a basic indexer? We want to use our engineering talent to solve more challenging and important problems. From the beginning the core technology has been coded in Java in order to interact well with others systems and let customers and partners to be able to work with out system and make customizations.
Do you have patents?
Ah, regarding intellectual property, I’m not sure this is a key issue as technologies are changing very quickly. But to answer your question, yes we are about to patent several of our system's methods. I will tell you more in a few months, okay?
Why search and content processing? Isn't there an easier software niche to pursue?
No, I don’t agree with you Stephen. First, I find this boiling market very exciting, because it is so competitive. This is what makes us get up in the morning!
Search and PolySpot are at the heart of the information challenge. The evolving needs of our clients make me think that this is just the early days… Un avant goût…
It is very gratifying when you complete a complex implementation and you see the smiles on the project managers' faces…They are playing with the solution as if they were a child with a new toy. This is really rewarding, an it makes information vital, not a source of dissatisfaction. Our system is engineered to allow employees to get their work not, not become a source of frustration.
You are providing a very point-and-click interface. What's the importance of the PolySpot interface?
Yes, point and click is an okay phrase, but I prefer simple and intuitive to use. Usability is key today. For too long the “large” vendors ignored user needs at this level and tried to brainwash the market with talk of algorithms.
We too have very clever algorithms, but it is important what the users see and how easy things are to use. The ease of use factor is just as important and to some of our customers, it's more important to have access to the information than know about mathematics.
PolySpot is one of a number of companies within the information access space, and whilst the underlying technologies may be able to access many, many sources, it’s our ability to present these effectively to the user.
This is where Information Access really comes into its own. With today’s technology users demand ease-of-use interfaces, and we have some delightful innovations coming into our interface very soon.
But remember we provide an infrastructure platform so our technology can also easily surface information within most enterprise business applications as well.
What technologies are you using to chop up, index, and manipulate the content you ingest into your system?
We have created our own content connectors and transformation modules. I think we are able to handle content transformation very cleverly. We can understand, extract and determine from the information a great deal of useful meta data. For example, we can use data from our transformation process to automate many processes such as document classification. These data are not then discarded. We use the information to tackle our customers' real business challenges. This is the key, we must develop our leading-edge technology to satisfy business challenges.
Can you give me an example?
Yes. Now we are working with a very large global oil field and information services company. This client is working with our engineers to develop one of our modules which we call “Trends”.
The idea is that our “Trends” function enables the client's employees to analyze and slice & dice information from across all of their call-centres. Any hot spot or important events which are important to the client's business will be easy to spot. We know that identifying trends of this type is important. We want to focus on making the one which may have a major impact across their business easy to spot.
Sure, “Trends” can be applied to many other business challenges everywhere but this is what I mean by developing technology to help such business challenges.
Let me follow up on that. Are you using knowledgebases or do you discover the information about the content you process on the fly?
That's a good question. PolySpot can do both. The typical model is to use our connectors to plug-in and aggregate information from knowledge bases. And when we do this knowledge process, we also map the security model and of course any additional fields of the knowledge bases. We maximize the value of metadata; for instance, we use these data for content classification and filters that mesh smoothly into any corporate security schema. Our knowledge base components are very enterprise ready.
Can you give me an example of real time processing in this on-the-fly function?
Yes, let me give you an example of “on-the-fly ?”. PolySpot contains a Document Collaboration module. You are in a research team for a large financial organization. You locate a useful analyst report about a company.
You can open the document and add a comment to it, appended to the original document. You can then put this into a public folder and forward it on to a colleague for his or her comments. We think this is like "document blogging" or annotating. These comments or additional information payloads are indexed “on-the-fly”. Other PolySpot functions can be used to make sure your colleagues have access to the information.
How do you determine how to balance the different processes so you deliver good relevance without choking the system?
That's a good question, Stephen. The concept of relevancy is very subjective. What can be relevant for one could not be for others. Freshness, type, tagged, source, category : all of these criteria are available for each user in order to ease him to find its own relevancy model, matching his or her context query.
So we let the users balance the processes to meet their specific needs. We don't like the "one size fits all" approach. People working in a company have too many things that change. The system has to make work easier, not put a busy employee in a digital straightjacket.
How will you position PolySpot in 2008 with the competitive ground shifting so rapidly? No one wants search. Now it's semantics or business intelligence.
We offer an infrastructure platform. If the client wants to think in terms of competitive intelligence, that's okay. If the client wants to process text to wring out knowledge beyond key word indexing, that's fine too. We have a system that allows us to solve business problems. We don't try to keep up with the marketing jargon coming from Madison Avenue or Silicon Valley.
We talk about the functions of our modules. We talk about using the system to resolve business problems. From this we have many exciting modules which can be easily applied to business challenges.
The beauty of PolySpot is that we are also so agile compared to the “big” companies, and we are determined to deliver the best “beyond search” platform in the marketplace.
Within PolySpot we have many engineer-years of experience in our team from the “big” vendors, people who have joined us from Autonomy, Verity, and IBM, among others. This market knowledge, our foresight from clients needs, and our agility help us to keep at the forefront of the market.
There's been a great deal of reluctance in the US to accept the fact that most users of search systems are dissatisfied with what their company makes available to them. What are you doing to help your customers keep their user satisfaction high?
You are correct. A lot of enterprise search customers using systems from the well-known vendors are not happy.
This is an end-product of the legacy vendors. That’s why we must listen to our users. I was glad to see that in your Beyond Search study you exposed these weaknesses and identified some of the vendors trying to address these issues.
You may recall when we first met, you mentioned the phrase "beyond search” to us and we were very happy, because this phrase was common to us. We have been using the phrase beyond search in our sales & marketing literature for quite a while. This idea is a big part of PolySpot's. Going beyond key words to information access is in the very soul of our company.
We try to listen to our customers. Even though we have made a sale, we want to keep the system doing the job the customer needs done.
We include a free annual usage analysis in our maintenance contract. This analysis looks at things like usage reports and try’s to see if the solution would have been misused or could be tuned better. Within a same company expectations from a group of people to another could be really different.
When you look at the trajectory of some promising companies that have crashed and burned what gives you confidence that PolySpot can continue its double digit growth?
I differ somewhat from other CEO’s. I am more focused on business, not the mathematics of our algorithms, which are very elegant. Before the PolySpot adventure, I was involved in several and different business experiences. From competitive intelligence to financial trading. I was also involved in many investments for software companies in the late 90’s both in the UK and the US.
I consider myself more like an entrepreneur. My philosophy is to invest wisely and carefully, both financially and with people. We have also been profitable for the last three years and this will continue. To be honest, I even expect that the double digit growth to continue. If we listen to our customers, we may grow even faster.
Are you relying on French-trained engineers or are you outsourcing some of your coding to other countries?
Right now we are 99 percent in house development. Next time you visit, we will go to a PolySpot lunch. Not only can you choose from many different cuisines near our office here in Paris, but you can have a multi-national lunch group. We are a true multi national company.
I think this is one of our greatest strengths. We find having different experiences and backgrounds a wonderful way to look at our customers' challenges. We are able to deeper experience and a greater depth of understanding of local requirements and needs.
We are committed to invest any of our profits in our research and development--both people and projects. This is important for us, not public relations talk.
When you look at the creative chaos in search coming from the acquisition of Fast Search by Microsoft, Engenium by MarshMcLennan, TripleHop by Oracle, what do you see in this market?
This is a difficult question. I am often asked, "What's going on? What's happening in search?" Maybe you are looking for an investment tip?
No, just your view. I am too old and tired to chase big money.
Okay, if you say so. But there is no easy answer. For sure this is still an evolving market. A good comparison of what could happen would be the consolidation which occurred on the CMS [content management system] market – more mature and very competitive- for the last five years.
Yes, much market consolidation has taken place which is an excellent business signal. Autonomy buys Verity who had previously acquired Ultraseek. Autonomy buys Cardiff and Zantaz. Microsoft buys Fast Search, who in turn bought Convera. This type of consolidation will continue.. But it is important for this market that independent companies like PolySpot survive. I think the smaller, more nimble firms are the ones best able to innovate right now. Look at the companies you included in your Beyond Search study. Most of those are smaller, very innovative, and very nimble.
Looking at 2008 and 2009, what are some of the trends that you see building now like waves away from the shoreline?
I will give you two for now.
Ok, the first wave should be using unstructured data to produce structured data. This is interesting, so far we have seen business intelligence vendors harnessing structured data, sometimes bundled with search vendors, to produce unstructured data.
To respond to this demand, we are about to release this business intelligence module based on unstructured data. At the end of the day our solution has to help end users to produce and react. As I suggested, searching for the sake of searching is not too productive. The BI trend has more utility, and we are right ion the crest this wave.
The second big trend is what we talked about when we were joking about wine and technology--usability. Not just for the person looking for information. No, we have to provide usability for the customer's own information technology team. The idea that a customer has to wait three or six months to get a fix for a problem is one of the weaknesses of the big, well-known systems. We are responding to this demand of our customers to control dynamically a user's work space, maybe a single person, maybe a team of two or three, maybe the whole subsidiary. I do not want to go into too much detail here but we have some very exciting developments coming later this year.
Without giving away any secrets, what's next for PolySpot?
We are making good progress as an independent company. It would be great if Google knocked on our door and made everyone in the company rich like so many Silicon Valley engineers.
Right now, it is not wise to mix Californian wine with French wine at the table. However good they may be. It may deliver some headaches for the next morning.
PolySpot's information access system delivers remarkable usability. In Beyond Search, the company was identified as one of the vendors to watch. With its focus on solving customer problems first and demonstrating what its system can do with real-world information access solutions, the company combines advanced content processing with ready-to-deploy functionality and easily-tailored interfaces. The company bundles connectors, access tools, and support for linguistic and statistical functions in one Java-based solution. This company's approach is refreshing in a sea of tool kits on offer from search companies with high brand recognition. PolySpot is worth a test drive and inclusion in a short list of behind-the-firewall search systems.
Stephen E. Arnold, May 19, 2008