(FORTUNE Small Business) – By 1996, Silicon Valley had spent nearly $1 billion trying to develop a handheld computer, and about all there was to show for it was a rather rotten Apple Newton. Then along came Donna Dubinsky and Jeff Hawkins. He glued together bits of mahogany and cardboard in his garage and transformed them into the first PalmPilot. But it was she who built up Palm Computing and turned the wooden experiment into one of the most successful consumer electronics product launches in history, selling 400,000 units during its 1996 debut.

Dubinsky repeated herself in 1998, when she and Hawkins left Palm--which by then had become a division of 3Com--to start Handspring Inc. Apparently running a small division within such a large company wasn't enough for the entrepreneurially minded 47-year-old. Venturing out on their own paid off. By undercutting Palm's prices by as much as 30%, Handspring's Visor gained market share quickly, representing one out of every four personal digital assistants sold after just a year in business.

Not bad for someone who almost traded her high-tech career for paint brushes. Unlike her Harvard MBA classmates, Dubinsky swore off white-shoe firms and took a $30,000-a-year customer-service job at Apple instead. Her title (and income) grew over the next few years, and eventually she headed up the company's software division, Claris. But Apple's unwillingness to spin Claris out sent her packing to Paris. "I learned a lot over there," she says. "Mostly that I was a lousy painter."

Painting may have some renewed appeal. The economy and Wall Street's disdain for all things tech have hammered Handspring. Revenues are down to $241 million this fiscal year, from $371 million, and at press time its stock price was at $1.50. But salvation may come from Treo, a new line of devices that packs a PDA, cellphone, and e-mail appliance into something about the size of a deck of cards. "It's the right product at the right time," says Dubinsky. Well, she's been right before. --Arlyn Tobias Gajilan

"Silicon Valley may have a big high-tech reputation, but in many ways it's a small town. I'd heard of Jeff Hawkins and Palm Computing through a friend who knew someone applying for a VP of engineering job there. But he was reluctant to pursue the position since Palm was still without a CEO and he didn't want to take a job without knowing who his boss would be. But the company and the CEO job interested me, so I asked Bruce Dunlevie, who was on Palm's board, to introduce me to Jeff. I got the job, and my friend's friend didn't.

The PalmPilot was born out of frustration. At first almost all we did at Palm Computing was sell handwriting-recognition software. But we got tired of seeing our great software run on others' so-so handheld computers. We'd tell them what we thought they could do to improve, but they wouldn't accept our little company's influence. Bruce Dunlevie got sick of Jeff and me complaining about our partners' lack of innovation and vision and said, 'You know, if you guys know what to do, why don't you go do it?' The next day Jeff shows up with a little cardboard and wooden model he'd glued together in his garage. That model became the PalmPilot.

But the eureka moment came when we showed the docking cradle and explained that our little device could connect to a PC with the touch of a button. Nobody had done that before. It seems basic now, but no one had made the logical leap that this organizer was a PC accessory, not a stand-alone PC.

I was excited to go out and sell it, even though bigger companies with deeper pockets than ours had already spent about $1 billion developing PDAs like Apple's Newtons, which didn't do well, in part because they were too complicated and expensive. When I set out to sell the idea to investors, I wasn't concerned about that history. But I guess I should have been, because the idea wasn't very salable. Basically, people had an enormous amount of skepticism at that point, and they just didn't buy it. Plus the timing was very bad. The Internet had just taken off, and people were very excited about that, not devices like ours.

It became very clear very quickly that we weren't going to get any VC backing. So we actually didn't spend that much time on it. We spent about a year trying to work with potential corporate investors and had better luck there. But each time we'd get into a negotiation with one of the big companies, we would face a morass of issues. Some had divisions that were feuding with other divisions. Some just had a very different vision of the future of handheld computing and PDAs, while others wanted certain rights and privileges that we just weren't willing to give up. For example, we worked with Compaq for a year trying to get funding from them. The idea was that they'd have a Compaq-labeled version, and we'd have a Palm-labeled version. We'd somehow coexist in the market. A lot of the discussion was about that coexistence. But we were very wary of being too dependent on a partner. We didn't want to lose control over our future. They wanted a clause that said we couldn't launch ours if they decided not to launch theirs. In effect, we would have been at their mercy. Still, that deal was absolutely tempting since we needed the money. Temptation aside, you just can't be starry-eyed and go into a deal and say, 'Oh, it'll all work out.' If the motivations are not aligned, then it won't work.

Ultimately we found the right partner in U.S. Robotics. At the time, they were the leading modem supplier. They looked like a potential supplier to us as well as a potential partner. So we approached them about an investment, and they came back and proposed an acquisition that made more sense for us. Up to that point everyone else wanted to integrate our technology into their product vision, essentially disassembling us and our product. But USR saw us as a new business that could run independently within their company. We were acquired by USR in 1995 for about $44 million.

We had an incredible debut year. We sold about 400,000 units, which was more than we could have hoped for. Keeping up with early demand wasn't easy, but we managed. Soon I started seeing our product in people's hands while they sat at airport terminals and at cafes. We were everywhere.

In 1997, 3Com bought USR and became our new bosses. From the start, I went to 3Com's CEO of the time, Eric Benhamou, and said, 'Listen, just put us off to the side and let us be independent.' We were really not on his radar at the time since we were so small. So he agreed. In the year that followed we tried to persuade 3Com management to spin out Palm and make it a truly independent business. But our success was making it harder for them to let us go.

It wasn't a good fit with 3Com from the beginning. We were under tremendous pressure to deliver results 3Com needed for its business. Every time we'd lower a price, we'd get nasty calls from corporate saying, 'You know, you can't lower your prices. You're going to hurt the gross margins.' It was true, but we needed to keep pricing comparable to that of our competitors. Plus 3Com was a company in trouble. They would do a quarterly expense forecast and ask for a 10% across-the-board expense cut. We'd say, 'But wait, we're growing and we're profitable. We're not the ones to cut.' They'd say, 'Hey, fair's fair. Everybody's gotta be onboard here.'

When 3Com finally decided it was not going to spin us out, I told Eric, 'Okay, well, then I have to leave because I don't believe that staying at 3Com is the way to make the Palm the most successful that it can be. Therefore I can't stay, and neither can Jeff.' So I quit for the both of us, even though Jeff wasn't in the room. Jeff was a little upset that I'd spoken for him, but he quickly agreed that it was the right thing to do.

I took three days off, and then Jeff and I got together and said, 'Okay, what next?' And we started creating Handspring. I set about doing the corporate things, like financing, finding office space, and negotiating with 3Com to license the Palm operating system to us. Jeff went to work on our next product.

Starting up the second time around was much easier. As it happens, I'd scheduled lunch with Bruce Dunlevie before I'd quit 3Com. He showed up at lunch with a check for $3 million. Then I went to Costco and bought Intuit's QuickBooks to set us up with accounting. Another friend, Guy Kawasaki, whom I knew from my days at Apple [he is now managing director of the business incubator Garage Technology Ventures], let us borrow space in his office, and then I rented three desks: one for me, one for Jeff, and a third just in case we hired someone else. I also bought a refrigerator. One of my favorite stories is when we'd done the first Palm, I had one of those little two-cubic-foot refrigerators. This time I bought a bigger one because I was feeling much more optimistic about our future.

We were primed, but we didn't yet have a product. Where we'd been the most innovative at Palm was on the high end. It hadn't shipped yet, but we'd pretty much finished the design of the Palm V and the Palm VII. We decided to focus on the low end and innovate and differentiate there.

We wanted something low cost, familiar, but still new. Jeff made a list of the things he wanted our new device to do, and it became clear that expandability was essential, but there was not a suitable expansion architecture at the time. Then he quickly outlined what became the Springboard expansion architecture. That enabled us to create a product that could become a phone, an MP3 player, a digital camera, and almost anything consumers might want.

We'd quit in July, and by October we'd hired the first engineers. It took us just 12 months from hiring the first engineers to the shipment of the first Handspring Visors. It's true that in a sense we'd become a PalmPilot clone company. We did have the expansion slot, but we remained a relatively familiar product. We wanted to do that. The big strategy at Handspring was to start with something known, something with low risk, something we could execute well too, and we did. Within 12 months we were back in the market with a differentiated product and captured 28% of the market. It was really one of the first alternatives to a PalmPilot.

When we left, Palm hadn't innovated that much on the low end of their product line. So when we started Handspring, we designed the Visor specifically with a low-end entry-price point that would be compelling to consumers. At the time the Palm VII was selling for about $450, while our Visor sold for $149. Palm owned the high end, and we didn't set out to compete for that space-- at least not initially. Around the same time, we started on our Communicator strategy, embodied by the Treo, which combines the functionality of an organizer with a built-in phone, web browser, and e-mail capability. Our line of organizers, meanwhile, was supposed to feed our new business.

The Treo began shipping earlier this year, not the easiest time to launch a new product. But we're confident that it's the right product at the right time. The idea for the Treo originally began with the Visorphone, a plug-in module for the old Visor. It launched in September 2000 and sold for $299. It may not have been very commercially successful, but it was extraordinarily successful from a learning perspective. Rather than just say we failed, we tried to understand what people liked and didn't like about the product. The Visor as a phone and communication device proved limited. People loved the functionality and they loved the utility, but they didn't like it as a plug-in module. It was clear to us that we were onto something. We said, 'Here's a direction, here's a trend.' We just absolutely believed that the future of handheld computing would be communications as much as computing. The Treo was our response to that realization. Sure, it was a departure from what we may have set out to do originally, but it was the right thing to do.

You have to be nimble. I doubt there's any entrepreneur who's succeeded on his original business plan. Situations change, the dynamic changes, the competitive environment changes, the technologies change. You have to be able to adapt.

You also have to be able to stop things. One of the disciplines I'm most proud of at Handspring is that we kill projects. That is so hard to do when people have invested their time and effort. And they're good projects. But you have to decide, Hey, when we started, it made a lot of sense. Today, for whatever reason, it doesn't. We can't keep doing it just because it's got momentum.

To keep your edge, you also have to get everyone who works for you invested and involved. Luckily, building a culture like that is very natural to us. Everything we do is always challenging and asking questions and probing things and trying to understand what the best course of action is. We don't pretend to know that, so we know the wider we cast a net in terms of getting that data, the more successful we'll be.

Doing that was easier when we started out as a 28-person company, but we're still doing it with 400 employees. We do lots of management by walking around. I do these things called Dine With Donna, where I get groups of employees together--sort of a group of a dozen or so--and get wonderful information about what the real issues are and what's going on in the company. I also have lunch in the cafe almost every day. It sounds like a stupid little thing, but to me it's one of the most incredibly important things I do. I could go off and have big honcho business lunches, but I sit out in our cafe because I can be with different employees and be accessible. That philosophy of easy access permeates the entire company."