Startup

by S. Jerrold Kaplan

  In the early spring of 1990, GO’s product development was heading for trouble. We had originally planned to release our first product by the end of March, but it had been repeatedly delayed, until the fall sounded more realistic.

  Part of the delay was due to IBM. Whenever its engineers met with State Farm, they would impose a new set of requirements on us. If I objected, Sue King would sometimes threaten to exercise IBM’s option to take the source code and walk away. This would have been fine with me, but only after IBM had helped to make us successful. We also faced the continuing problem of raising another round of financing. The IBM agreement would be of little use in doing so unless someone there was willing to confirm its existence to prospective investors.

  The solution to all this seemed obvious: make a press announcement. Nothing focuses an engineer’s attention like being pestered by friends and family to show them the product, and this process begins in earnest when they first read about the product in the paper. Once IBM went public with its support, it would be much harder for them to back away from us. And it would be a lot easier for GO to raise money with a pile of glowing press clippings.

  The first step was to agree on a date for the announcement, so Quinlan took the matter up with IBM’s PR department. It seemed that the department’s main function was to prevent the release of information, not to hasten it. The publicists insisted that the announcement of our relationship not conflict with anything else in their schedule. Given the number of pending announcements—most of them matters such as releasing new models of disk drives and awarding grants to charities—the earliest they could allow was July 18, a long way off. The silver lining was that this would give us time to properly brief the press.

  When the PR staff at IBM learned that Quinlan had assured us that Jim Cannavino would make the official announcement, they tried to prevent this from happening. To Quinlan’s embarrassment, Cannavino was suddenly unavailable, leaving Quinlan to handle the event himself. “Cannavino will take calls from the press, however,” Quinlan said in consolation.

  The next stumbling block was the ritual press release. IBM assigned a young publicist to write a first draft, which he did, but without bothering to contact us first. It was a nightmare, proclaiming that IBM had acquired rights to a new technology from GO that would help bring its own handwriting-recognition research to market more quickly. As it happened, he was authorized to draft the release but not to change it. For a company that grew prosperous on data entry machines, it was odd that IBM restrained its own people from using the delete key.

  “If you’d like to discuss this, I’ll fly out and we can go over it together,” the publicist offered. Instead, I was able to persuade Quinlan to start over with our own draft. Even then, the publicist tried to stick paragraphs from his original version back into the release.

  Carol Broadbent, our newly hired director of marketing, who was in charge of corporate communications, worked diligently with GO’s PR firm to position the announcement in the best possible light. This was a tricky matter, as we weren’t actually announcing a product but only a “relationship” and a “technology,” things the press was justifiably skeptical about. Carol shuttled me around the country on an endless publicity tour, filling every available hour with interviews at important publications.

  One day in mid-April, Carol stormed into my office with a fresh copy of PC Week. “Take a look at this!” She waited, fuming, while I scanned the headlines. Sure enough, there was a front-page story announcing that IBM had made a major investment in GO and had licensed its handwriting-recognition software to us—neither of which was true. PC Week, a particularly aggressive publication, must have somehow gotten hold of IBM’s early draft of the release. “They should rename it ‘PC Leak,”’ Carol said.

  I called the reporter and politely inquired as to why she hadn’t checked with us in advance. “You’d just deny it,” she explained. It must have been a slow week.

  Carol was frantic, and with good reason: we had sprung a leak long before an announcement—a PR person’s worst nightmare. We immediately set up a command center to do damage control.

  Our first line of defense was to try to contain the story. Carol sequestered me in an office and lined up calls to everyone we had previously briefed, asking them to continue to respect our embargo. It helped a lot that the published story was so inaccurate.

  Emerging from the office late in the afternoon, I walked down the hall to give Carol a final report on my assignment. “OK, that’s about it,” I said. “Everyone’s covered, and they’re all going to hold the line.” She was visibly relieved.

  When I got to my office the next morning, she stormed in again, brandishing some rolled newspapers as though about to smack a dog for soiling the rug. “Read ’em and weep,” she said, passing the roll to me at arm’s length. Both the New York Times and the Wall Street Journal had run follow-up stories—without contacting us, and sidestepping their responsibility for accuracy by prefacing the news with the disclaimer “PC Week yesterday reported . . .”

  “I can’t believe this would interest them,” I said in disbelief.

  “Can’t even do their own research,” she added. “No one called to check with us. It’ll be worse tomorrow.”

  “Why?”

  “It’s on the wire.” She was referring to the Associated Press, which fed news continuously to its subscribers via teletype. The AP was a key source of material for papers around the country.

  Carol was right. The story was picked up by hundreds of local newspapers the next day, often edited to dispense with the qualifier about PC Week in order to improve readability.

  We were both demoralized. “What should we do next?” I asked.

  “You better get out there and make some more news,” she said.

  As urgent as this seemed, there was the more pressing problem of raising money. We had managed to stretch our cash for a few more months, and with the additional IBM money, had enough to last through June. Given the delay in the announcement, Quinlan had kindly agreed to talk to selected investors on our behalf to confirm the IBM agreement. I pressed John Doerr into service.

  “I’ll ask our existing investors for their pro rata,” he said, “but we’ll still need a lead to set the price.” He was referring to the common practice in which existing investors take part in later rounds in proportion to their percentage ownership, but only if a new investor is willing to set a price and participate.

  I knew what to do. My next call was to Norm Vincent at State Farm. After all, this is how insurers made some of their money, by investing the premiums they collect in advance each year and earning as much as they can on the “float” until they have to pay the money back in claims. Also, State Farm was well aware that the IBM deal was for real.

  “Norm, this is Jerry. I just wanted you to know that we are about to raise another round of financing, in case State Farm might be interested in investing.”

  “I’ll call our investment department,” he said without hesitation.

  If there was one group at State Farm that I expected to be different from the others, it was the investment department. I imagined a fast-paced trading floor in a Wall Street high-rise, with nervous M.B.A.s sitting at banks of computer screens, watching every move of the financial markets. I was wrong. Soon I was back at headquarters in Bloomington, but on a different floor with a different mission. The investment department looked as quiet and reserved as a bank. A single computer display sat unattended in a corner, apparently tracking State Farm’s $60 billion in investments. I made a brief presentation, then went to lunch at a country club with Norm Vincent and two investment officers, where they inquired about the asking price.

  “Our investors are willing to go at two-fifty per share,” I responded. This was true, but perhaps a bit misleading. Kleiner Perkins was willing to support any price that I could persuade a new investor to pay. After due consideration, State Farm had tentatively committed $5 million to help us out—at the $2.50 p
rice, a dramatic markup from 75 cents in the previous round. This placed the company’s value at close to $75 million, a high figure for GO’s stage of development. But I had my lead, and everyone else was willing to follow.

  Then John Doerr arranged for us to visit Andy Grove, the CEO of Intel, as a possible investor. Intel had a near monopoly on the CPUs—the central processing units—that are the heart of DOS-and Windows-based personal computers. An outspoken Hungarian, Grove ran his company with the intensity of a field marshal. I went to Santa Clara and demonstrated our product to him and several lieutenants.

  Grove focused on the prototype. “What microprocessor does it use?”

  “The Intel 286,” I responded proudly.

  He pursed his lips in disappointment. “Why not the 386?”

  “Your salespeople didn’t tell us about it in time.” This was something that really annoyed me. After convincing us to stick with the 286, Intel started putting up billboards all over the place showing a sign painter crossing out “286” and writing “386” over it in red paint. Intel was trying to persuade everyone to upgrade to its latest, greatest product. I had even written a letter of complaint to Intel’s VP of marketing. Fortunately, no one remembered the letter.

  “We’re always interested in good software for our processors, and that’s why we’d be investing. Are you planning on porting to any other processor line?” Grove asked. He knew that we were as much a danger as an opportunity. With a new operating system, we had the option of switching to a competitor’s chips if we wanted to; then, if we were successful, he would be forced to play catch-up.

  “Not at this time,” I responded guardedly.

  “We’ll bet five million—but with the understanding that you’ll upgrade to the 386 as soon as possible and agree that new versions of your software won’t be released on competitors’ CPUs before Intel’s.”

  I looked at John. He was trying to nod his head vigorously in agreement without being too obvious. I thought Grove’s proposition was probably OK, but others were investing without conditions. “I don’t know if we want to saddle ourselves with these restrictions for the same price that others will pay without special concessions.”

  “And just who might you be thinking of porting to?” Grove asked with eyebrows raised. John clutched the arms of his chair, worried that I might blow the deal. “Look, Intel isn’t like the other investors. When we announce that we’re backing you, it will really help your market momentum.” This was true, and a reasonable argument. Besides, I couldn’t imagine GO redesigning the pen computer around another CPU, so I agreed to his conditions.

  Grove assigned one of his staff to negotiate the contract, and we worked up a detailed legal agreement covering the restrictions. Whenever I sounded hesitant, the negotiator would remind me of how valuable Intel’s public support would be. And when we were about to close the deal, he called me up to request a change. “Jerry, we’ve just had a meeting about the investment, and we’ve changed our mind about a few details.”

  “Like what? I thought we were done.”

  “We’re only willing to invest three million, not five. And we don’t want to announce the investment publicly. In fact, we want it in the agreement that you won’t tell anyone at all, except for a few investors and other partners. And only under an NDA.”

  I was quite irritated. “What is this bullshit?”

  “This is what we’re willing to do. We can revisit the announcement question again after you upgrade your system to the 386.”

  “I’ll call you back.” I hung up the phone and called John Doerr.

  “John, Intel’s trying to jerk us around. They cut the investment to three mil, and won’t announce.”

  “Calm down, calm down,” John said. “I think Grove may be in the middle of negotiations with Bill Gates, so he’s probably sensitive about rocking the boat right now. It’s just good business.” He thought for a moment. “We need the money. Let’s do it anyway.”

  He was right. Replacing the $3 million would have been difficult, and the other investors were expecting Intel to participate. Reluctantly, I called back Grove’s negotiator and agreed.

  With State Farm and Intel spoken for, it was a simple matter to round up the rest of the money. By mid-June we closed on $15.31 million, just in time to pay the bills for the announcement with IBM.

  “Sue King’s very concerned about your schedules and the recent changes to your user interface.” Mike Quinlan sounded as if he were letting me know I had bad breath right before a blind date. “She’s recommending that we hold up the announcement.”

  I called Sue directly to find out what the problem was.

  “You added icons to the system without telling us,” she said. “Now it looks a lot more like OS/2, so that group is arguing once again to kill the project and go with them instead.”

  At first I couldn’t imagine what she was talking about. Then I realized what it was: after careful study, our user-interface design team had enhanced the “notebook metaphor” to allow for multiple notebooks. These appeared as tiny pictures of books at the bottom of the screen. When you wanted to open one, you tapped on it. This was only superficially related to the icons in Windows and OS/2, but that was enough to threaten the entire IBM relationship yet again. I had to think fast.

  “Sue, would you please tell the OS/2 people that those aren’t icons? It’s a bookshelf.”

  “What?”

  “That’s part of the metaphor. It’s a bookshelf where the notebooks are kept. Besides, it’s across the bottom of the screen, not along the right side like in OS/2.”

  She took this argument back to fight off the OS/2 group and soon dropped the request to delay the announcement.

  Finally, the announcement week arrived. IBMers showed up at GO two days early for the purpose of reviewing what everyone was going to say. Quinlan brought along a new IBM face, a woman named Kathy Vieth, who he had previously informed me would be joining him to help promote the project both inside and outside IBM. Then Quinlan took me aside to tell me some bad news, which he had a habit of holding back until the last minute: after the announcement, he was “retiring” and Kathy would henceforth be GO’s primary contact.

  The last thing we needed at that point was another changing of the guard, particularly since it was clear from watching them that Kathy Vieth and Sue King didn’t get along very well. After working to develop a relationship with Quinlan, I wasn’t looking forward to starting from scratch again. But the show had to go on.

  Despite the premature disclosure in PC Week, Carol Broadbent had managed to pack the room we rented at a nearby hotel with more than seventy reporters by promising them more news and free food—I’m not sure which was the more potent draw. After the crowd picked over the cheese and fruit, the moment for our first public announcement was at hand.

  The staging was spare, like a Shakespearean set, except for a gleaming silver construction of our newly redesigned logo, illuminated by a spotlight: two interlocking circles that spelled GO in script, both right side up and upside down, enclosed in a blue oval. Our new VP of marketing, recently stolen away from Steve Jobs at Next, had insisted on paying $70,000 to a high-priced design firm for this blue egg. “It’s a bargain,” the marketing man had said. “Next paid a hundred grand for their logo.” He wasn’t with the company long.

  I was punchy for lack of sleep. As Carol dimmed the lights, I leaned over to Quinlan and whispered, “This is your last chance to back out.” Fortunately, he laughed.

  I got up and approached the podium to speak. “I’m pleased to have this opportunity to announce to you a new partnership between GO and IBM in the area of pen-based computing.” Suddenly the lights went out, strobe lights started flashing, and warbling sirens wailed so loudly that people covered their ears.

  The fire alarm had gone off. It was impossible to continue, and the hotel refused to shut off the alarm until the fire department arrived. I stood at the podium for several minutes, making hand signals to the
audience to stay put. I considered doing the announcement in pantomime.

  Of course, it was a false alarm. By the time the mayhem ceased, the audience was thoroughly distracted and partly deaf. I needed to regain their attention. “I knew this announcement was hot, but not that hot!” The room exploded in laughter. A number of reporters used the incident as the lead for their articles.

  When I was through, Quinlan gave a convincing speech about IBM’s unwavering commitment. Then Kathy Vieth got up and seconded Quinlan’s remarks.

  After the event, I instructed my executive assistant to send Quinlan an appropriate gift. Quinlan called me up the next day. “Thanks for the beautiful bouquet,” he said.

  “Oh shit, is that what she sent?”

  “Don’t worry, my wife isn’t the jealous type.”

  Despite the earlier leak, the press was kind to us, giving us frontpage coverage in most computer publications. Except for PC Week—they buried the story on page 135.

  The press had been unfair to John Sculley at that time, portraying him as a manager with little vision, a mere caretaker for Steve Jobs’s Macintosh legacy. To improve his image, he was looking around at Apple for an exciting technology that he could take ownership of. He picked the pen computer project, by now code-named Newton, which had languished since Steve Sakoman’s departure. He asked Larry Tesler, his technology guru, to take charge.

  Sakoman, a hardware type, had focused much of his attention on a secret joint-development project with AT&T’s Bell Labs. AT&T had custom designed to his specifications a hot new type of low-power but high-performance microprocessor known as a RISC chip, called the Hobbit. There was only one other similar chip available, called the ARM, for Acorn RISC Machine, offered by a British technology company.