An Interview with Oleg Shilovitsky
After a series of meetings in Boston last week, I met Oleg Shilovitsky, publisher of an interesting Web log PLM Think Tank and the Beyond PLM Web log. I had learned about Mr. Shilovitsky’s stealth start up Inforbix and, after months of sleuthing, convinced him to meet with me at Thai Basil on Newberry Street. What interested me about Mr. Shilovitsky’s venture was its applicability to manufacturing and production. Most enterprise search and content processing systems tackle email, standard office documents created in Word, and data locked in enterprise software systems like Lotus Notes or SAP’s aging R/3 platforms. Not Mr. Shilovitsky. His focus is mission critical information that most vendors ignore or cannot process with their established methods.
Let me give an example. In a small manufacturing company, a traditional search system would index the content on an Exchange server, email, proposals in Word files, and maybe some of the content residing in specialized systems used for accounts payable or inventory. When these items are indexed, most are displayed in a hit list like a Google results page or in a point-and-click interface with hot links to documents that may or may not be related to the user’s immediate business information need.
But what about the specific part needed for a motor assembly. How does one locate the drawing? Where are the data about the item’s mean time before failure? The semantic relationships between bits of product data data located in multiple silos are missing. The context of information related to components in a manufacturing and production process is either ignored, not indexable, or presented as a meaningless item number and a numerical value.
That’s the problem Mr. Shilovitsky and his team of engineers has solved. With basic key word retrieval now a commodity, specialized problems exist. As Mr. Shilovitsky told me, “I think maybe we have solved a problem for the first time. We make manufacturing and production related data available in context.
The full text of my interview appears below:
What's your "angle" on search and content processing?
I think, the landscape of content processing and search for businesses and enterprises will go through the period of significant changes. There are multiple reasons for that. First, there is the influence of the consumer market and the Internet. Next is the growing amount of information in the organization. Third are customer demands for efficient decision making. And, finally, most organizations are putting more cost pressure on information technology units.
My belief is that companies need to re-think the way they are working with data. Today manufacturing companies accumulated huge amount of data. However, most of this data cannot be used into an everyday decision process by people. Hugging data is another syndrome. Part of the "rethinking process" will be a new understanding of openness and value of information.
How did you get interested in this particular facet of search and content processing?
I saw many problems related to information access in manufacturing and production operations. The idea for Inforbix emerged over the last decade working for Smart Solutions, SmarTeam, Enovia, Dassault Systems. Over these years, I held various positions in R&D group and company management. Most recently, I was the ENOVIA SmarTeam’s Chief Technology Officer.
What’s your background?
My background roots are actually in computer software for industry design and automation. In my university time (back in USSR of 1980s), computer means a separate room. Later in 1990s PC came, and I discovered a wonderful world of CAD systems. Since then, I have been working in the domain of engineering and manufacturing software. I've been involved into development of multiple product data management tools and commercial systems for the past 20 years in companies in Israel and elsewhere.
What type of system is required to crack the information retrieval challenge in this domain?
In my view, the most valuable characteristics of future systems will be "flexibility" and "granularity". The diversity of data in manufacturing organization is huge. You need to be flexible to be able to crack the information retrieval. On the other side, businesses are driven by values and ROI. So, to be able to have a granular solution (don't boil the ocean) in order to address a particular business problem is a second important thing.
What are the salient features of the system you have developed?
Our system foundation combines flexibility and granularity with a deep understanding of prodcut data in engineering and manufacturing. One of the problems of product development is a uniqueness of organizational processes. Every organization runs their engineering and development shop differently. They are using the same tools (CAD, CAM, CAE, data management tools, or an ERP system), but the combination is unique.
Our deep belief is that "one size doesn't fit all" in manufacturing companies. Our secret sauce is to combine flexibility, granularity and understanding of product data semantics. Another part of our secret sauce is to work on features with your customers. This is why we have decided to become visible and open very early in our company development. So, many of our product features are in discussion between our development team and our current and future design partners.
What differentiates your approach from others that are available to prospective licensees?
That’s a very important question. I can identify few important characteristics: We deliver cross silo data access. We also have developed a proprietary and highly precise method for product data semantics support. Efficiency is very important in today’s manufacturing and production settings, so we have focused on a low implementation effort. Unlike a one-size-fits-all search system, we have designed modules and can deliver quite granular applications.
The combination of these characteristics will create an important differentiation between our solution and alternatives that available for our customers these days.
Could I use your system for eDiscovery or a similar activity where I don't know exactly what's in the corpus?
Our fundamental application assumption is that a user doesn't know what information he needs. How can an engineer ask a question without having some prior knowledge about a new problem. Most users will only have a hint or some related information such as the name of the customer, project, or an assembly. This information cannot drive him to the actual product data directly. However, indirectly user can follow a system of links and semantic references that help him to discover data.
What's the licensing approach you offer your customers?
Inforbix has currently a development project. We are working with our design partners to figure out what will be the best licensing approach. I suggest we will get back to these in the future interviews.
Isn't the search and content processing sector crowded? I track about 300 firms and there are more contacting me each week.
I think, content oriented sector is stepping into a very interesting phase - understanding of what means "one size doesn't fit all". Google is good for the internet. Some other vendors are suitable legal information discovery. The industry is moving towards significant vertical orientation, in my view.
Will you highlight a use case that catches the attention of some of your prospects?
Our customers are diving into product related information. However, the data and information are scattered between networks, files, applications and other sources. One of the most interesting use cases for us is to find connected pieces. Our users want to consume product data as a whole, when it comes to decision making. You can follow our blog for so called "product data stories". We are planning to share our future experience there.
Do you support other vendors’ systems or are you a stand-alone solution?
We are dependent on data that located in other systems. There are files, databases and other systems. We are working selectively with files and systems that drive interests of our first customers.
What's your view on semantics and natural language processing? Are these technologies ready for prime time?
I think semantics and language processing are important. However, I can see lots of much simpler cases related to content processes and data analyses. I think usage of semantics, and natural language processing will gradually come to different fields.
A number of vendors have shown me very fancy interfaces. The interfaces take center stage and the information within the interface gets pushed to the background. Are we entering an era of eye candy instead of results that are relevant to the user?
I can see two interesting trends. People like everything visual. At the same time, simple always wins. You can tell me, probably, that these two are not going to the same direction. However, this is actually, the biggest challenge - to make it user experience visual and simple.
What are the hot trends in search for the next 12 to 24 months?
I think we will see more vertical solutions trying to get more precise and dedicated way to analyze and consume data.
Where can people get more information about your system?
Inforbix project is open to the world of social networks. We are already open Linked, Facebook and Twitter account. Together with Product Data Spaces Blog on http://inforbix.com, we will provide all available information about what we are doing. Just following us online for more information.
General purpose search vendors are revamping their products and services for niche markets. The most common examples are companies like Attensity pushing into advertising and marketing services for the Madison Avenue crowd and OpenText acquiring content management vendors in the firm’s effort to address problems in information production.
Inforbix is a specialist firm with a next generation search system that solves a problem in a key business sector: manufacturing. Like Linguamatics in the UK, Inforbix offers a solution that addresses a real world problem. As a result, the value proposition is easy to communicate. Access to manufacturing information reduces costs and improves the overall product process. Costly and time consuming duplication of effort can be eliminated. In my experience, flawed information management means that engineers may design a common part multiple times because the previous design and drawing were easily accessible. The phrase “just do another one” is a profit killer and I have heard it many times in my career.
Companies like Inforbix, in my opinion, have a bright future. A generalist indexing system cannot provide the fused data elements needed for manufacturing. This is a company to watch.
Stephen E. Arnold, November 3, 2010