Columns

From AI to applications, Intellicorp shifts with the times

• By Michael W. Bucken
• July 5, 2001

Marty Hollander was there when it all started. He saw Mountain View-based Intellicorp. carry the dream of artificial intelligence (AI) forward, left to pursue other endeavors, watched as AI fell into disfavor and then rejoined the company just as it reformulated itself as an enabler of SAP applications development. Editor Mike Bucken recently spoke with Hollander, now Intellicorp vice president of marketing and business development, to consider the company's prospects.

Can you give a history of Intellicorp?

The company was formed around 1980 as part of an effort that was going on at Stanford University to apply Artificial Intelligence technology to the new field of genetic engineering as it was coming about.

And out of that, the team that was doing it realized that not only did they have tools for genetic engineering, which is probably completely unrelated to what the company does now, but the core technology really sprung the basis for the beginning of the expert systems industry. So the first product was what keyed Knowledge Engineering Environment.

And it incorporated some very interesting techniques that are becoming more mainstream today. The one that's best known, of course, is object-oriented programming. Intellicorp became a leading expert systems company and grew, but the market changed, and the company, I think, remained focused over a decade on building development environments to improve people's productivity in building applications. Not that there's anything wrong with that, but the company continued to serve the market for people who build applications as opposed to solving end-user problems.

Why and when did the company shift gears to focus on applications?

It was a combination of things. SAP and Intellicorp had actually been working together for now about five years, so probably sometime around 1993.

It was in 1995 that the major change occurred where the company realized we shouldn't just be playing with SAP; we should actually grow the business around it. And it was also determined at the same time SAP started to get the pressure to shorten implementation time as R/3 became a richer and richer environment and more broadly used.

Have you been able to take advantage of AI technology?

I think we should look at what makes up AI technology. A piece of it has to do with using object-oriented technology as your basis for storing model objects because that allows you to do a lot of things with them that are really important for a robust modeling environment with a repository and a multiuser setting.

So I would say that core experience of being knowledgeable on how to use object technology has given a huge advantage in being able to grow the functionality of the product from a modeling point of view.

The second major area has to do with what would be referred to as truth maintenance systems and the use of constraints. And this is really playing a part in the next stage in what we're building, which is called the Configurator. There are lots of different configuration engines used for different things.

When you're dealing with a model and you can make choices about turning things on or off or utilizing different components and getting rid of certain other ones, given there are so many inter-relationships in the SAP software, you need to ensure that those decisions that are made by users do not have an impact that would cause something to be inconsistent elsewhere in the system.

So we have worked with them to build and utilize the technology, Configurator, which allows them to be confident that changes made in one area will be correctly reflected, or notify other implementation teams about changes that one group is making that could have impact upon another.

What is the strategy of the purchase of the Universal Portable Interface from Deloitte and Touche?

I do think it's important to look at application integration and how that fits in with business process modeling. Our insight is that our partners need to do application integration as a major component of the design step and the implementation stage. Live Model helps them on the design stage of looking at where the interface points are between R/3 and these other applications.

What we decided is that it's natural for us then to provide tools that enable them to actually do the integration of those once they've done the design.

We looked around, and we said we're going after the SAP space, which is the major vendor. We found the best product for integrating applications into R/3 and that was the UPI product from Deloitte and Touche. So we purchased the UPI product line.

Can you say something about UPI?

It's about 15 people right now, and they had over 100 customers. In the application integration space, there was no other company that had that size customer base, that number of successful users.

There has been a four-year development effort of continuing to grow and improve the UPI product in the context of R/3. It is solely an R/3 product. It was both a cash and shares purchase. The value was around $7 million.

Who are the competitors here?

That's a very competitive market. The one that's most famous is a company called CrossWorlds, previously known as CrossRoads. They've raised almost $40 million. They believe package-to-package integration is the focus.

We've learned that legacy integration is an extremely important part of the market. We are focused in on both with a real strength on the legacy integration side. Other players range -- you have the old line players like Information Builders or Evolutionary Technology TSI, who are moving into the space from having data integration tools.

As SAP continues growing, do you expect increased competition?

In the modeling space, it's interesting. To really play in this space you need to be approved by SAP. If you can't get the reference model licensed from SAP then you can't really author a tool in this space. And there's only a few licensees.

Second, since we are building SAP's next generation business engineering environment with them, Intellicorp has a huge inside edge, in terms of understanding where SAP is going and what it needs and what the customers require in order to do business engineering for R3. So there are other competitors.

The one that is most well known is IDS Scheer out of Germany, which built the first generation product. We are building the third. SAP tried to build the second.

The others are sort of viewing tools -- for example MicroGraphics with ABC Flowcharter. They have just announced a product called Enterprise Charter. I don't really see them as competitors but really alternative ways to have these models on your desktop for viewing and annotating.

What about plans for the Web?

We've built a Web-based product for SAP -- Business Engineer -- it was released with 4.0a that SAP just came out with, which is totally built-in Java and it's a viewer that runs basically inside of a browser and you can view these models. So, for just viewing, you can get from SAP as part of the purchase of the R3 system a way of viewing those models from any Web browser. That's the first viewing component we built for SAP.

How do you describe Intellicorp today?

Well, it is a completely changed company from the past. Clearly, you can see our revenue growth for six consecutive quarters. The last three quarters have seen increasing profits. I think what you're seeing is a company focused in on customer problems as opposed to technology.

We've learned how to build extremely strong partnerships, both with the software companies like SAP, as well as with consulting firms who need to utilize these tools, like Deloitte and Touche and IBM Global Services. We're now putting in place the components for building a long-term successful company.

What convinced you to return to Intellicorp?

It's sort of like a child. I came back after being gone for a decade because this company could have been on its deathbed. I always love a challenge. Can we make this company be vibrant again? That's what we've done.

About my second week here, I took on SAP as my project. I didn't know what SAP was, to be honest. But I understood we had a very big company that had a problem that was looking to work with us and we needed to find a market for this company.

And that was also at the same time that we needed to move from being a Unix company because that's really what we were. The challenge was we were not only a technology company but we were focused in on Unix, and we needed to move our development environment and our whole base of code to the PC.

And rather than port, we built from scratch in the context of solving SAP's
problem.

Will it be easier to build tools for other packaged apps?

It will be very easy because what changes is the content. And the content is supplied by that software vendor. The tools can work with another application with almost no modification.

The only modification is really the connection to that software product's system for the simulation part -- the ability for the models to actually connect to the transactions and to be able to fire those off and display the screens as it's simulating, walking through the models. And that's probably the only unique piece. The repository, the Live Model, visualization tools are all consistent across packages.

How long can you concentrate on SAP only?

The market is growing. We've barely penetrated the market. We believe that the complementary software packages that integrate with R/3 are a very viable next step for us in terms of market expansion. And that's a very big market.

Do you expect Intellicorp to become a standard modeling tool for SAP?

Well, the question is how do standards come about. I believe that standards come about through use by companies or individuals that find them valuable.

And I believe from a business process modeling view, SAP is finding that these models are core to their vision of how to implement packaged applications. I think many other people will move along in that same way. Whether it has to be a standard that the whole world agrees upon, I'm not sure. It would be nice if that occurred, but I don't think it's required.

The beauty of what we've done is we stored as a model, so we understand the semantics of the methodology. If we need to convert to something else, we could actually do that conversion and display another methodology as long as it contained all of the nouns and verbs of that language.

So we don't have to be religious about what's the right way to display it or state it. It just needs to be able to describe business processes and the decisions that are made. That said, I still believe that more people are using the EPC methodology for business processes than any other methodology.

We hope that both through becoming the standard on SAP and hopefully related packages and being used by the majority of the world's implementation teams to do packaged application integration, that, to some extent, we are the de facto standard. So we don't need to be approved by some standards body. I think it's through actual market penetration that standards are truly developed. And I think we're moving down that path.