Kim Polese was one of the driving forces behind Java, the programming language that brings the Web to life. Now with her company, Marimba, she's setting in motion another Net revolution.
New software companies seem to be popping up everywhere these days. Some arrive with much fanfare but end up delivering little. Others have clever technology but lack the marketing skills to become a success. A third, much rarer type is able to combine impressive technology with an ability to grab attention. California-based Marimba is one of these companies.

At Marimba's helm is Kim Polese. A former employee of Sun Microsystems, Polese was responsible for launching the Java programming language in early 1995. With a virtually non-existent marketing budget, she propelled Java to the forefront of the computing world. From the relative obscurity of Sun's development labs, Java has become one of the computing world's most talked-about technologies in just 18 months.

Marimba's other three founders are also ex-Sun employees and Java experts: Arthur van Hoff, Sami Shaio and Jonathan Payne were all members of the original Java development team.

The reason for the sudden interest in Marimba is the company's first product, Castanet. Castanet offers a new method of delivering information across the Internet, overcoming its inherent problems of limited capacity and slow transmission speeds. Since its launch in mid-1995, Castanet has caught the attention of industry heavyweights such as Microsoft and Netscape and has a worldwide user base of 100,000, and growing.

Industry watchers believe Castanet has the potential to transform the way people use the Internet. In a sector where product hype is rampant and every new release is hailed as ground-breaking, that's a big statement, but one the company's founders are confident will prove true.

"Marimba is delivering something so big and so compelling that I think it actually does match all the build-up and hype," Polese says.

The hype has been considerable. In May last year analysts linked a dramatic dip in Sun's stock price to a short Wall Street Journal item about key staff leaving to set up a new company - Marimba. In September, media interest grew further when the US venture capital firm Kleiner Perkins Caufield & Byers announced that one of the first beneficiaries of its $100m Java Fund would be Marimba. With so many eyes on the company and so much interest being expressed by established industry players, it is easy to forget the company is just 11 months old.

For Kim Polese, establishing Marimba was a natural step. "It was the realisation we had that there would never be a more perfect time to get out there," she says. "We also had the visibility from being key [Java] developers and that really catapulted us in the first few months."

Polese's impressive career path is intrinsically linked with Java. She joined Sun Microsystems in 1989, and soon became product manager for the then new programming language C++. Polese had learned the language during her university studies but found that she enjoyed the marketing side of the business more than generating code. "The fact that I had a technical background allowed me to make the transition into marketing more smoothly," she says.

Polese's big move at Sun came in 1993 when she became involved in a project to develop the computer language that was to become the precursor to Java. Called Oak, it was designed to run within interactive televisions and simple hand-held devices. Although versatile, the high cost of devices running the code meant it failed to take off. Consumers simply weren't interested.

Around this time Polese, together with others at Sun, realised that the Internet, rather than interactive television, was going to be the next major IT growth area. It was decided this should be the focus for the language. Everything that made Oak perfect for the set-top box environment also made it perfect for the Internet. The code's simplicity, compact size and adaptability made computer networks its natural home.

Together with a team of developers, Polese devised a marketing plan for Oak, now renamed Java, which positioned it for desktop PCs and the Internet. In May 1995, the code was put out on to the World Wide Web for the first time and Java usage exploded. "The technology was picked up very rapidly by developers and companies who recognised its power," Polese says.

But despite a huge initial response, Polese still had a big job to do. This included educating those within Sun about Java and its potential impact and responding to thousands of queries from developers and users.

"I also focused on developing some 'real-world' applications built using Java," she says. "It's one thing to explain the capabilities of the technology in theory, but it's another to be able to say, 'Go and check out what Morgan Stanley has built with it'. There's no doubt, though, that the job was made a lot easier by the fact that this really was killer technology. The world had been waiting for it without realising it had been waiting."

With her name strongly linked to Java, Kim Polese was now firmly on the map. Offers of venture capital came in to enable her to start her own company. While keen to do just that, Polese and her team held back until they had developed a prototype of Castanet. Once complete, each of the four founders invested $15,000 and established an office. Marimba began trading.

The key advantage of Castanet technology is that it allows information to be transmined, or "pushed", across the Internet rather than users having to search it out. Once a user subscribes to a Castanet "channel", updates can be received regularly and stored on the computer's hard drive via Marimba's software "tuner". This ensures the user has access to up-to- date information without the frustration of long waits for downloads.

The number and diversity of companies using Castanet is growing daily. The car manufacturer Chrysler is preparing to install Castanet tuners in each of its showrooms. The tuners will give customers access to real- time information on delivery and manufacturing schedules.

The PointCast Network, which uses the Internet to provide constantly updated news and information to subscribers, is considering using Castanet as a delivery system. Apple will bundle the Castanet tuner with its new Macintosh operating system; Microsoft is evaluating Castanet and may consider bundling it with its Internet Explorer, while MGM is to use Castanet to broadcast soap operas from its Web sites.

A US-based organisation is establishing a channel to provide fast, up- to-date information about missing children. "The problem with putting up photos of all the missing children on a Web site is that it takes forever to download," explains Polese. "Castanet overcomes this problem. Also, if you use a channel, you can personalise what is received. For example, users can learn about children who have gone missing in their area."

It is this ability to personalise information sent to users that makes Marimba's technology of particular interest to certain companies. A retailer, for example, could provide customers with individual catalogues, based on personal preferences and likes. "It finally makes the Internet a viable way to do business," Polese says. As with any new development in IT, other companies are examining Marimba's technology and rushing to produce competing products.

"A lot of companies have emerged in this area of 'push' technology, but we are different," Polese says. "Not only are we solving the problems of data distribution, but also the harder problem of software distribution and updating."

This latter area is where Marimba and Castanet shine. Many companies are becoming increasingly concerned about the cost of updating software for users. Distribution via the Internet is attractive but until now the sophisticated technology required has not existed. The difficulty stems from managing versions of software. "When you introduce a new piece of code, you have to make sure that it doesn't conflict with something that is already there," Polese explains. "It becomes a much more complex problem

Kim Polese is confident Marimba has solved the challenge of software distribution via networks and can deliver a solution in the form of Castanet. "We are going to be a major player in the Internet space and I believe we will be shipping a standard for the way that people distribute software and update their applications," she says. "Technically, we have really figured this out."

Some industry analysts have positioned Castanet as a competitor for Microsoft's ActiveX suite of software. Microsoft does not agree with this comparison. "ActiveX encompasses a whole range of components," says Mike Pryke-Smith, Microsoft's Internet tools product manager.

"The core base of ActiveX technologies is now in the hands of an industry standards body, the Open Group. It's no longer in the Microsoft domain. Also, there is a whole industry that exists around ActiveX already. It is not something that we have just brought out. The technology it uses has been around now for about five years."

One analyst predicted Microsoft could come up with a Marimba-style product within about six months if it devoted sufficient programming resources to the project. But the feeling is that the company is more likely to work with the newcomer rather than against it. Why reinvent the wheel?

This spirit of co-operation is already in evidence with the recent announcement that the browser giant Netscape is to integrate the Castanet tuner into its new component technology.

Kim Polese is brimming with enthusiasm about the new relationship. "Users won't have to worry about software, just about getting their jobs done," she said. "Software, data and files will simply be there, personalised for them, whenever they plug into the network."

Perhaps Marimba may succeed where other companies have yet to truly deliver - making technology so simple to use that it becomes secondary to the content. If this is the case the message may at last become more important than the mediumn