Silicon Valley is full of bright young technology company start-ups that blaze one day and are gone the next, usually when their One Big Chance falls through. But Be, a small software company that many people wrote off 16 months ago when it was wooed, then rejected by Apple Computer, has shown the kind of tenacity against all odds that earns industry respect and support.

And that's despite or even perhaps, because of the fact that Be makes only one product: an operating system. Acting as a kind of internal foreman controlling all operations, the OS is the most fundamental piece of software in a computer, the soul of the machine.

Creating a new one is a monumental, risky task but at the end of last month, Be released the first commercial version of its OS (there were two developer prereleases). About the same time, Be received an influx of $27 million in additional venture capital, indicating continuing investor support for a task many may consider mad, if noble.

Operating systems other examples are DOS, Windows 95, Windows NT, the Macintosh OS, IBM's OS/2, UNIX and Linux are few and far between, difficult to develop and slow to mature. One reason is that computer users like being able to run the software they already have, which enables them to use all the data they've stored in machines over time, so most operating systems have to incorporate "legacy code", the instructions used in earlier, related systems.

Often, it's outdated and clumsy and isn't particularly good at making use of the growing capabilities of the hardware. Software developers hate having to work around its limitations. New operating systems also take time to mature the accepted industry belief is that an OS needs about 10 years before the bugs are worked out, it becomes stable and its in-built capabilities are maximised.

Be's founder and chief executive, Mr Jean-Louis Gassee, believes the BeOS, after nearly eight years of development, has reached a sufficient level of maturity to be offered to the world as a commercial operating system, not just a developer's plaything. Why now? "At the end, it's a judgment call which has to do with solidity, depth, some feel for critical mass," he says. "It's for the public to agree or disagree."

So far, the public, at least in the form of the crucial software developers who are needed to back an OS with applications, has been enthusiastic, if cautious (while far less visible in Europe, Be has followers in Britain and Mr Gassee's native France, where Be maintains its European office, says Mr Gassee). The BeOS has gathered about it a growing circle of developers now numbering more than 5,500

who muster the kind of excitement about Be which hasn't been seen since Apple launched the Macintosh.

That's a process Mr Gassee is familiar with one of the brighter lights at Apple Computer in its early days and a prime mover and shaker behind the Mac, he has always managed to retain a useful (in Silicon Valley-speak, leveragable) amount of street credibility. So many developers were willing to have a look when Be unveiled the first versions of its OS. Many were won over instantly by two compelling features: it's a spanking new OS "clean", as developers say with no clumsy legacy code. It's also very lean and simple its 700,000 lines of code make the BeOS the Kate Moss of operating systems. In comparison, Microsoft's high-end OS, Windows NT, contains 25 million lines of code.

"That kind of muscle it's like an athlete. If you lift weights, you're not going to be a good dancer," says Mr Gassee. NT "is definitely not a dancer it's a sumo wrestler". Litheness is an important feature for the market Mr Gassee is trying to reach with his OS, the media crowd who work with audio, video and other graphics files, all big-memory items. The BeOS is, says Mr Gassee, "the media OS".

For a while 18 months ago, it had looked as if the BeOS might instead be the Mac OS. As the Mac had grown older and in-house development seemed incapable of redesigning the Mac OS with some key contemporary features, Apple Computer had gone shopping. The BeOS seemed a natural fit with Mr Gassee as its parent, it has a similar feel to the Mac OS and had the features Apple wanted, "multi-threading" and "pre-emptive multi-tasking" (both of which make the operating system more stable and efficient). The buyout came so close to conclusion that the two main Macintosh magazines, MacWorld and MacWeek, ran covers in January 1997 touting the BeOS as "the new Mac OS". But as Mr Gassee reportedly asked a price at which Apple was balking, Apple's founder, Mr Steve Jobs, gatecrashed the party and managed to convince Apple to buy his own company, NeXT, instead.

For a while, Be management could be stinging about Apple at developer events; Mr Gassee is now publicly supportive of Apple but begins to bristle if asked too many questions about that phase of Be's history.

"I don't want to live in the past when before us is a huge opportunity," he snaps, but adds bitingly, "The Apple space with 2.8 per cent of market share has no interest to us."

To the surprise of many, Be has instead moved into the vast Wintel (Windows/Intel) space with much vigour a development which certainly must have helped the company when talking to the venture capitalists of the Valley.

But Be works in both Mac's PowerPC and Windows environments, one of its most interesting features. Rather than present the BeOS as a replacement OS for a machine's existing operating system, the slimline BeOS actually runs alongside either Mac or Windows, creating a partition for itself on the hard drive. Documents and other items can be passed back and forth between the two operating systems, and the user can switch between the two systems with ease.

That feature may be the BeOS's key ticket to success - a 66-person company can hardly challenge what Mr Gassee calls "the Windows fortress". Instead, it can pursue "a nice, peaceful coexistence with Windows". The Wintel space is "geek heaven", he says, "both huge and ecumenical" (the sometimes strident battle between the devotees of the Mac and Windows platforms is often referred to as a religious war).

So Mr Gassee is in the unusual position of having an OS which can be marketed almost like a software application for a niche media market rather than as an operating system an approach which has never been taken before. "I don't know if we will ever be a mainstream OS but I think a geek market is a very good landing spot," he says. "Our goal is to have a very small but very profitable portion of the market." He says Be, while still working off venture capital, should reach profitability sometime next year.

So small is beautiful? "When people ask if I'm satisfied to be a niche player, I say, `Damn right'."