Be started out around 1991 by ex-Apple employees (including Apple executive Jean-Louis Gassée and Newton's inventor Steve Sakoman -- who now is back at Apple managing the iPod division). Some ex-NeXT and ex-Apple engineers and Amiga developers joined Be later on too. The "father" of BeOS (in terms of the original OS code), Benoit Schillings (these days working at Openwave), joined the company around 1992-93 and started working on many parts of the OS. Legend says that BeOS was only text-mode back then and the company needed money. It was Friday night and JLG asked Benoit to put together something "fancy and graphical" before Monday in order to show it to venture capitals and impress them. And that's how the first app-server for BeOS (a graphical server like X11 or GDI+ or Quartz) got created: in under 3 days.
Mipsys, Raphael Moll and later Beatware were the first people among those who received BeBoxen (the first two got the H0bb1t version, while the Motorola PPC-based BeBoxen started shipping after 1995). "The OS was "BeOS 0.99 exp", codenamed "shark", it was regularly updated via a couple of floppies. The next year we received the first blue PPC BeBox, though I can't remember the exact date. The OS came on a CD; I had been told we were the first ones to receive it" Raphael remembers.
Here is the only known screenshot of the "shark" version of BeOS, 0.99-EXP from 1994 running on a H0bb1t machine (which used 6-7 AT&T DSPs instead of regular CPUs). Around the time I joined BeNews.com in 1999, the site published a great timeline article presenting all BeOS releases from 1994 until 1999, but because BeNews is now down indefinately you can read the english text on web.archive.org (note: it's very slow but it does load) and check out the screenshots at a Lithuanian site which has translated that article.
While I personally started using BeOS on March 2nd 1999 (thanks to the PCPlus printed UK magazine for including a live CD with the R4 version) I later met a lot of people who had been using the BeOS already for years. It is kind of romantic hearing all these stories, for example, a developer who later became a Be engineer had to carry his BeBox to his house from the post office in his arms (and the BeBox was a very heavy machine compared to PCs), my husband (who was actually one of the first Be developers worldwide and later also became a Be engineer) spending crazy amounts of time with his BeBox in 1996 instead of... studying or sleeping, etc.
I probably owe BeOS some gratitude myself: I was arguing about some tech details of the nvidia TNT2-Ultra card and its BeOS 2D driver in Feb 2000 on the BeNews forums with a Be engineer. Later, we took our disagreement over email and then moved it on IRC, on the #BeDev channel. A year later, we were married and we've been happy ever since. Today, JBQ works at Openwave along with 4 more ex-Be engineers.
Back on topic, here is another story I love: It was 1997-8 when Dominic Giampaolo (from SGI, later worked at QNX and now he is at Apple) was testing his baby, the 64-bit BFS file system which earned him lots of fame for BeOS and for himself. There was a specific QA stress test that would use BFS to write on a floppy disk, erase, rewrite, erase etc for a whole night. For some reason, according to the debugging logs, the write procedure would bail out always around 6 AM. Engineers would gather and analyze the problem, but no one could figure out why the test would always bail out at around 6 AM, every morning. So, Dominic decided to stay up all night and have a watch at the machine personally. What Dominic found as the culprit was really funny: apparently, a ray of sun light would enter the window and would fall directly on the floppy drive and that would cause the drive itself to fail for the duration the sun light was upon it!
And here's another story: Baron Arnold (the person that "owns" your files on the single-user but somewhat POSIX-compliant BeOS if you do an "ls -l" on your files) was known to be able to find bugs... easily. He never followed pre-defined QA methods but used his intuition and his skills to find bugs during his time as the main QA person at Be. Many ex-Be engineers still remember Baron finding a bug on the BeBox itself: inside the Motorola CPUs. Motorola even flew an engineer at Be to track down the problem. He earned having his name on Motorola's errata! Today, Baron and a bunch of other ex-Be engineers are working at Danger.
Here's another kinda-funny one: It was early 1999 and Be was preparing the R4.5 release. BeOS, which was targetted to be a MacOS replacement around 1994 naturally had to resemble MacOS a bit on usability (so it could get more switchers). So, to do a copy/paste you had to use ALT+C/V, which are the Mac shortcuts for the operation. However, in 1999 BeOS was not targetting the Mac anymore, but the PC market. Many PC users found the ALT+C/V --instead of CNTRL+C/V-- very annoying and were becoming vocal about it. JLG asked Dominic and some other engineers to add the functionality to be user-selectable of which shortcut to be used. Half of the Be engineers didn't even want to hear about it because the BeOS architecture was not able to allow this as an option. JLG pretty much had to pull rank and ordered the engineers, in an angry manner, to pull this through. Finally, it was implemented and to this day, it remains a hack. Legend says that if you were outside the Be building that day, you could hear a lot of kicking and screaming...
Disagreements on how to implement proper multi-user functionality were also present, mostly between kernel engineers, Dominic and Pavel Cisler, the creator of the BeOS desktop/filemanager, Tracker. Pavel came from General Magic and later worked at Easel developing Gnome's Nautilus while today he works for Apple on Finder. Half of the engineers were citing kernel/fs changes and other half filemanager ones. Multi-user functionality was finally implemented but was never shipped because it was breaking a lot of apps that were created with single-user in mind (BeOS already had about 1500 applications at that point, today it has about 3,200) and that was a business risk Be didn't want to take.
For those who loved the platform, remembering these and many other stories that I won't mention here, it leaves a taste of melancholy: "aaah, those were the days..."
Anyway, later, Be published the last major version of BeOS, 5.0 in March 2000. A year and a half later Be's assets were purchased by Palm and its source code and engineers moved to PalmSource, Inc. PalmSource made it clear that they have no plans to utilize the BeOS in any fashion, but a German company created by ex-Be third party developers, YellowTAB, is aiming to release its BeOS-based OS, Zeta, later this year. Zeta is based on the unreleased updated code of what it would be a "BeOS 6" and it is currently on version 1.0-RC3 pending release. More info here for their latest news (in Japanese, but shots are provided).
The BeOS legacy might live on via the Zeta product and/or OpenBeOS, however it will never feel the same as it used to feel in the 4.5.2 days (according to many engineers, the best version of BeOS ever released -- for its time). The OS just felt like it had a soul, like it would know what you were thinking when using it (even if BeOS does have its own technical problems). It felt pure. I am not using BeOS anymore (I boot to it once every 1-2 months or so) but I will always keep with me this feeling, a feeling that no other software ever given me.