Gil Syswerda, "Founding Startups"

USE THESE BUTTONS TO SHARE THIS EPISODE:

Angel Invest Boston is sponsored by Peter Fasse, top life science patent attorney.

AI wiz, founder and angel Gil Syswerda on building algorithms and companies. This episode focuses on Gil’s biography and his startups. It’s the first in-studio interview since spring 2020. We had a blast! Click here to listen to part two of this interview.

Click here for full episode transcript.

Highlights:

• Sal Daher Introduces Gil Syswerda, Founder and Super Angel

• Gil Syswerda Lucked into a Great High School – Built His Own Calculator

• "Computer programming. What is that?"

• “Because programming a computer for the first time you have this machine that just does whatever you tell it to do. The power is just indescribable. It just seems like the possibilities are unlimited.”

• Selling Cutco Knives to Pay the Rent – Learned to Sell

• “...I taught myself programming on that computer at Calvin College...”

• “When I wasn't at a place working, I would just fly to a city where I had friends and hang out with them. I just literally lived nowhere for a year.”

• Gil Meets His Wife Carol - They’ve Been Married 38 Years Now

• “One day I was at my job as a programmer, the next day I'm in class.”

• Animation Toolkit Was a Flop Because It Was Too Much of a Niche Product

• “The target audience is miniscule. We just didn't realize it. It was the classic mistake. If we build it, they will come.”

• Genetic Algorithms: A Lesson from Evolutionary Biology

• Gil Syswerda Gets into Hot Water with His Advisor by Being a Know It All

• BBN Labs – Gil Syswerda Wrote Seminal Papers in AI that Are Still Frequently Cited

• “What I really wanted was a job where I could use genetic algorithms and the only place on the planet that I could find that did anything of the kind was BBN Labs...”

• “I tried every way I could think of to get a ride in an F-14 but I never succeeded.”

• “Just because you're the CEO doesn't mean you have complete decision authority.”

• Optimax Systems Was Bootstrapped by the Founders – Had Revenue Right Away

• Raised Money to Cover Working Capital – Michael Mark Invested

• Pitch for Pater Fasse, Patent Attorney at Fish & Richardson

• “...created a machine for making scheduling systems and user manuals.”

• “We did not want to try to perpetuate the engagement with a customer by doing consulting...”

• Optimax Systems Followed Its Business Plan; Never Pivoted

• Shell Oil’s Tank Farm Broke Opitmax’s Optimization Algorithm

• Built a Scheduling System for the Mayo Clinic, Made the Mistake of Thinking Mass General Would Buy It Too

• “It was a very rapid four-year run to get an acquisition offer from i2 Technologies. We beat them in every single sales situation.”

• Optimax’s Revenues Had a Direct Positive Impact on i2 Technologies’ Earnings

• What Gil Syswerda Did After Leaving i2 – Travel the World + Race Cars

• The Big Lesson for Startups from Being a Race Car Driver

• Gil Syswerda’s Management Workflow at Optimax Systems

• Gil Syswerda Believes in Putting Boards to Work

• “...instead of becoming a professional race car driver, I decided to go back to my roots and build another hammer. Again, it's based on genetic algorithms...”

• “Machine Insight was to correct a really big tactical error that we made with Optimax Systems.”

• “I didn't even have to finish the presentation and they're already sold.”

• More Gil Syswerda to Come in Future Episodes

Founding Startups

Guest: Super angel and founder Gil Syswerda

Sal Daher: This podcast is brought to you by Peter Fasse, patent attorney at Fish and Richardson. 

Sal Daher Introduces Gil Syswerda, Founder and Super Angel

Welcome to Angel Invest Boston, conversations with Boston's most interesting angels and founders. Today we are privileged to have Gil Syswerda with us.

Gil Syswerda: Sal, it's a pleasure to be here.

Sal Daher: Gil Syswerda is both a repeat founder, very successful founder, and a very successful angel investor, and a colleague at Walnut Ventures. This is also a very special podcast in addition because Gil and I are actually in the same room. I haven't done a face-to-face podcast since sometime in 2020, so it's been more than a year. 

I'm actually in a very special place to be face-to-face. I am at PodSpot at Venture X in Marlborough. PodSpot is a podcasting wonderland. It is a studio set up by my friend and sound engineer, Raul Rosa, and he hopes to have many more PodSpots. There are going to be videos of this and you'll be able to get a sense of what this place is like.

We thought it'd be fun to have Gil here, and there's going to be video of Gil, video of and Raul in this gorgeous place that Raul set up here. Anyway, let's launch into this. Gil is an artificial intelligence maven. He's made his money in artificial intelligence. He studied at University of Michigan and he's the kind of guy that Michael Mark, my mentor in angel investing, this is a guy who with his 25-year-old professor from MIT, I think Michael was maybe 23 or something, started first company. He's just off the charts brilliant. He thinks Gil Syswerda is one of the most brilliant people around. Just get an idea who we're talking to here.

Gil Syswerda: That's great because I think Michael is one of the best investors around.

Sal Daher: Yes, he is. Of course, Gil is turning bright red. He's the type that turns bright red, very humble about this. We have somebody who thinks very deeply and Gil always asks very perspicacious questions in our Walnut meetings. Anyway, let's start talking, just a quick run-through through your biography. Tell me a little bit where you started from.

Gil Syswerda: I'm going to start pretty far back and I'm just going to rapidly roll forward. The kind of companies I've done have been high-tech companies.

Sal Daher: Before that, where did you grow up?

Gil Syswerda: I grew up in Grand Rapids, Michigan or Granville, Michigan. I was a pretty geeky kid. Geeky in a good way. The geeks who like to learn things and build things. There's a nerdy component to that as well. I was probably nerdy too. The things that interested me back then were electronics. I got an early start in computer logic circuits because of a kit somebody bought me and so on. I've built a lot of that kind of stuff.

Sal Daher: Was there entrepreneurship in your family?

Gil Syswerda: None.

Sal Daher: What did your dad and your mom do?

Gil Syswerda: My dad was a factory worker and my mom was a nurse. They came over from Europe, the Netherlands. I was actually conceived in the Netherlands and born here.

Sal Daher: You're an immigrant too.

Gil Syswerda: Technically not, but I don't mind being called an immigrant.

Sal Daher: No, no, you were conceived over there and you came over here.

Gil Syswerda: Legally, it's where you were born.

Sal Daher: That's right. Biologically an immigrant. Awesome. University of Michigan. When did it dawn on you that you wanted to be in the software area?

Gil Syswerda Lucked into a Great High School – Built His Own Calculator

Gil Syswerda: Let's back up just a little bit because the University of Michigan is kind of far in the future. When I went to high school, I actually ended up going to a pretty good high school just because of circumstance with my parents. I was allowed into this high school. It had amazing science labs and so on, and great teachers. It was a Christian high school but the religion part didn't stick with me, but certainly, the technology part did. One project I did, I don't know if you remember, but it just made a big impact on me is that I used to read Scientific American and one day there was an ad for Bowmar calculator. I don't know if you remember this thing.

Sal Daher: I remember the name, yes.

Gil Syswerda: It was a four-function calculator, LED display, and at the time it was just jaw-dropping. I can't remember ever wanting anything as badly as I wanted that calculator.

Sal Daher: This would have been what year?

Gil Syswerda: This had been maybe 1972.

Sal Daher: I remember circa 1967 and what was a calculator, that would have been five years before, was an HP device that was the size of several bread boxes and it sat on a desk. My dad was at MIT at this time. He was taking an engineering class. They would set up all the equations and then he would book time on the calculator. It did exponents and things like that, and then he would do all the evaluation, in that 45-minute period. That was a calculator circa 1967.

Gil Syswerda: It was probably really expensive.

Sal Daher: Oh, yes. MIT had 20 or 30 of these in this particular department and you had to book time. You know what I mean? You couldn't have one at home.

Gil Syswerda: Remind me, I'll bring up a Wang calculator that I used. It's a little bit later.

Sal Daher: Although I think this was a Wang calculator. It wasn't an HP, it was a Wang calculator.

Gil Syswerda: The one I use was a Wang 720C as I recall.

Sal Daher: Was it about this big?

Gil Syswerda: It was about the size of a big typewriter. It had two rows of Nixie tube displays, and it had a keyboard. It wasn't a keyboard. It was just all these chiclet keys that every function you could do in that calculator was programmable. It had associated keys, a lot of keys on this thing. I looked up the price of that thing. In today's dollars it would cost over $40,000.

Sal Daher: Oh, yes? Unbelievable. Unbelievable. The next step down from that was the slide rule.

Gil Syswerda: Well, yes, that's what we used.

Sal Daher: Yes, I had a slide rule in high school, then college that's when I got my HP-35. That was really first calculator.

Gil Syswerda: Yes, same track. In high school, we were using slide rules and then this calculator is in Scientific American.

Sal Daher: The Bowmar.

Gil Syswerda: The Bowmar. All right, the Bowmar Brain. In today's prices that calculator cost $1,600.

Sal Daher: Wow.

Gil Syswerda: It may as well cost a million dollars. There's no way I was going to buy that thing. But at the lab in the high school and they had a kit there that somebody must have donated. It was brand new. Nobody had even opened it. It was a breadboarding system. You could plug in integrated circuits and then lots of little wires that you could stick in holes so it connects things up. I decided to make my own calculator.

I studied all these chips that were like AND gates and OR gates, JK flip-flops, counters, pulse generators. All individual little integrated circuits. It took me a long time because I just had to figure it all out from scratch, but in the end, I had a device, hundreds of wires and only a two-digit display because that's all there was, and it could add, subtract, multiply, and divide. I was pretty proud of the thing.

Sal Daher: This is why Michael Mark told me that the guy was just off the charts. [laughs] You figured out how to build your own calculator.

"Computer programming. What is that?"

Gil Syswerda: Anyway, that was a memorable thing I did. A more important thing I did, and again, thanks to the high school, is they had vocational week for seniors. What that was is the seniors got to spend a week at someplace learning the beginnings of vocation. There were all kinds of choices. You could spend a week with a baker or with an accountant or whatever. One of them was computer programming, and we go, "Computer programming. What is that?"

[laughter]

Gil Syswerda: We'd heard of computers, not programming. Three of us decided to take this thing. It was an elective at college. Here's how naive we were. We got together and tried to figure out what is a computer? What's our knowledge of a computer? Our reference point was an article in Time Magazine where they referred to computers as giant electronic brains. They had a picture of boxes in a room and tape drives and so on.

Sal Daher: This is slightly a [unintelligible 00:08:31].

“Because programming a computer for the first time you have this machine that just does whatever you tell it to do. The power is just indescribable. It just seems like the possibilities are unlimited.”

Gil Syswerda: That was pretty good. Our other reference was Star Trek where Captain Kirk could talk to a computer and ask a question. He'd ask very Google-like questions. We literally sat down and made up a list of questions just in case we could ask this giant electronic brain a bunch of questions. Then in the end we went through the class, they taught us how to program in BASIC, and a teletype. The school had a mainframe computer with teletypes and it changed my life. Because programming a computer for the first time you have this machine that just does whatever you tell it to do. The power is just indescribable. It just seems like the possibilities are unlimited.

Anyway, then I ended up going to the college that had that computer. It was a combined program of three years at Calvin College, two years at University of Michigan engineering track, and I got a degree at the end of it. I started out in architecture, and then I switched to core engineering, then I switched to electronics engineering, but I just pretty much spent all my time with the computers that they had there. Including that Wang computer, but also the mainframe. I just got really good at that stuff. I wrote my first machine learning program. I wrote a program that learned how to play tic-tac-toe by playing itself.

Selling Cutco Knives to Pay the Rent – Learned to Sell

[laughter] It was just one of these things that you just did, and there was almost nobody to even tell about it because nobody did computer stuff back then, not in my circles anyway. I did do one thing in a summer that really impacted pretty much my abilities to do startups. I got a summer job selling Cutco knives.

Sal Daher: Oh, Cutco knives. [laughs]

Gil Syswerda: Cutco knives.

Sal Daher: If you happen to have a kid in the family who decides that he or she is going to sell Cutco knives, these are very expensive. They actually work pretty well.

Gil Syswerda: I still have some of those knives.

Sal Daher: Oh, I have drawers full of Cutco knives, but, boy, are they painfully expensive.

Gil Syswerda: They're very expensive, and you had to go out and sell these things to--

Sal Daher: All your relatives, to all your poor, poor relatives.

Gil Syswerda: You hit the relatives, and then, you sclhep door to door. I had to pay the rent. I finally figured out something. This would probably only work in Grand Rapids, but a lot of people were immigrants from Europe, and young ladies put together hope chests.

Sal Daher: Oh, hope chest with Cutco knives.

Gil Syswerda: What I would do is, at noon, I would go to downtown Grand Rapids. Everybody's out for lunch here, pick a young woman approximately the right age for a hope chest and explained about the knives and so on, and get an invitation to demo them. Of course, when I arrived at their house, she's living with her parents so, who would be standing on the front porch but the father greeting me. I would then have to sell the knives to the father while the daughter is listening in, and try to convince him to spend what the equivalent today would be thousands of dollars for a set of knives.

Sal Daher: What were some of the tests that you did to display how much better the Cutco knives--? "Stay away, the Cutco knives cut."

Gil Syswerda: I cut paper, cut leather.

Sal Daher: A tomato?

Gil Syswerda: I cut a penny in half-

Sal Daher: Oh, yes.

[laughter]

Gil Syswerda: -with the scissors that came in there. Here's the thing, I got really good at it. I sold a lot of knives. First of all, whatever shyness I may have had at that point, was gone because you cannot be shy and do this kind of thing. I learned how to sell in pretty difficult circumstances with really disapproving fathers-

Sal Daher: Oh, I can imagine.

Gil Syswerda: -and then end up selling thousands and thousands of dollar’s worth of knives. Anyway, that probably played a big role in my ability to just deal with sales situations even though my background was really pretty technical. It's a very useful thing to have. After Calvin College, I didn't really want to be an engineer. It just didn't resonate with me and, so I thought I'd take a little bit of time off and become a programmer. That extended into a few years. I ended up working for a company where I was the general troubleshooter.

Sal Daher: How is it that you re-skilled as a programmer?

“...I taught myself programming on that computer at Calvin College...”

Gil Syswerda: Well, I taught myself programming on that computer at Calvin College, so I was just pretty much self-taught. Then, I talked my way into programming jobs and just learned on the job and so on. I didn't have a formal education in it, but I did fine. That was, by the way, the days of air travel cards. Do you remember that?

Sal Daher: Yes, like travel miles?

Gil Syswerda: No, you could get a card that will let you travel for free, no additional expense, on any domestic airline.

Sal Daher: Oh, yes.

“When I wasn't at a place working, I would just fly to a city where I had friends and hang out with them. I just literally lived nowhere for a year.”

Gil Syswerda: There was a year where I gave up my apartment, put everything in storage and lived nowhere. All I had was that air travel card. When I wasn't at a place working, I would just fly to a city where I had friends and hang out with them. I just literally lived nowhere for a year. Then, I got tired of it, so I asked for a more permanent assignment and I ended up at Ann Arbor, Michigan, where there was a big deployment going on for EDS.

Sal Daher: For what?

Gil Syswerda: EDS, Electronic Data Systems. Ross Perot.

Sal Daher: Ross Perot. He was a big government contractor.

Gil Meets His Wife Carol - They’ve Been Married 38 Years Now

Gil Syswerda: This was actually a government job but it was located pretty much on North Campus of University of Michigan. I got an apartment just off of North Campus. As luck would have it, across the hall was an apartment full of University of Michigan co-eds, including a cutie named Carol. This Friday is our 38th anniversary.

Sal Daher: Oh, congratulations.

Gil Syswerda: That worked out. We had two kids and a granddaughter now.

Sal Daher: Awesome.

Gil Syswerda: It was a fortuitous move. I was in Ann Arbor. I didn't really like my job anymore and I thought, "I'm going to go back to school." Carol was pretty convincing, "You have to go finish things up." I had met some professors already because I was reading some of their papers and so on. This is now three weeks before classes start. I go to the admissions office and try to get in and they laughed at me because it's three weeks, four classes are starting. I decided months and months ago. I went to those professors, got letters of recommendation.

I went back to admissions and they let me in as a candidate for non-degree. The admissions person was very stern on me too. She says, "You're not part of a degree program. Your counseling is not going to be available to you. You can take whatever class you want. Nobody is going to tell you whether that's the right choice or not." I'm thinking, "This was a problem?" 

Sal Daher: Total freedom!

“One day I was at my job as a programmer, the next day I'm in class.”

Gil Syswerda: I gave notice. One day I was at my job as a programmer, the next day I'm in class. My first class was a graduate-level class of the philosophy of language and mind. I'm sitting there in that class. I'm listening to the lecture and I realize I've just majorly improved my life. I started studying computer science for real. There were a whole lot of terms that I didn't know like data structures and so on. I took every AI class. I got to know the professors. I met some people. I wasn't but more than a year into it, and three of us decided, I'm not sure even how we came about that, we were going to start a company, a software company.

Three college kids have just decided to do this. We decided just because of our interest that we should do 3D CAD type software for the IBM PC. All the CAD systems are running on mini computers and so on, expensive. We wanted to build a cheaper one here for the PC, which had very primitive graphics, but had some. We did that for a little bit working on the software. Then the Apple Macintosh came up. I remember the first time we saw it. We went down to the local store, and there was the original Macintosh. It had a 128k, that little black and white screen, but it was graphics, and it had a mouse. I thought, "Wow, we're going to have to switch."

Sal Daher: If you've been in Boston, Ralph Wagner would have been selling you that computer.

Gil Syswerda: Is that right?

Sal Daher: He had a Mac store.

Gil Syswerda: Oh, he did?

Sal Daher: Yes. Ralph Wagner, Flunk Calculus, Ace Life is the name of the episode. I did an interview with him. It was hilarious.

Gil Syswerda: I'm going to have to listen to that.

Sal Daher: Oh, you got to listen to it. It's highly expurgated because there's a lot of stuff that Ralph said that could not be made public.

Gil Syswerda: Oh, Ralph has great tories.

Sal Daher: It could not be sent out into the waves. Raul is here. By the way, Raul, say hi.

Roll Rossa: Hey, everybody.

Sal Daher: This is a part of the secret of my success is Raul Rosa, my sound engineer. That's why we sound so great. Roll is in the room here because we're checking out PodSpot the first time. Anyway, you saw this, first time, Macintosh.

Gil Syswerda: Macintosh, got smitten by it. They were really expensive. In today's dollars, that itty bitty little Mac with one floppy drive cost $6,000. You couldn't program it. To program that you had to buy a Lisa, which is a biggish Mac. In today's dollars, the Lisa cost $25,000, maybe a little bit more. You had to buy a Macintosh and you had to buy the Lisa. They were both really quite slow. Then to just develop software, it was really quite a painful thing and very expensive. We raised money.

Animation Toolkit Was A Flop Because It Was Too Much of a Niche Product

We eventually came up with a product called Animation Toolkit, which allowed you to make little animations on the Mac, like GIFs today. Nothing like that existed in the past. We had some very nice packaging done by the company across the hall from us that was into 3D graphics. Then we tried to sell it. It's not like today where you can go to Amazon and buy something. You had to run ads in magazines, or you had to get it positioned in stores. We were barely past the age where you went to a computer store and the software was on a floppy disk with a couple of Xerox pieces of papers stuck in a plastic bag and with a hole punched through it.

Sal Daher: Michael Marks does.

Gil Syswerda: Yes, Michael Marks does. How do you get your software into stores nationwide? We realized that we had made really a serious mistake, a business mistake, completely naive. We had produced a product that we thought we'd like to use ourselves while we were college kids.

Sal Daher: Classic mistake.

Gil Syswerda: It was cool and all that and you get some articles written up on it, but it wasn't quite a game. It didn't have staying power for the home user. It wasn't useful enough for business. Worst of all, it was priced too cheap because we needed to hook up with a distributor and a distributor has expenses. They want to make a profit margin, and our price was just simply too low for distributors to make money. It's a pretty naive business mistake. We did end up selling a bunch of them and kept the lights on and so on, but it wasn't a big hit.

Sal Daher: It's funny. I'm right now listening to Tom Eisenmann's, he's a professor of Harvard Business School, Why Startups Fail. This is a classic and it's in there. [chuckles] I'm preparing for interview in the future. He's going to be interviewed on the podcast. This idea that it's something that you feel the pain point and you think that you are a very typical customer. The reality is there are very few like you, especially like you, Gil, or someone you like, Michael Mark. How many are there in the country?

“The target audience is miniscule. We just didn't realize it. It was the classic mistake. If we build it, they will come.”

Gil Syswerda: Exactly. The target audience is miniscule. We just didn't realize it. It was the classic mistake. If we build it, they will come. That was just something wrong. There were bigger plans to build a word processor but I decided I was going to back out of that whole thing or at least spend less time on it and go back to school and study AI. By then, John Holland was my advisor for master's degree. John Holland is a pretty famous guy in the computer science circles. He invented the field of genetic algorithms. I can explain a little bit with that--

Sal Daher: Yes. Please explain that. I was privileged to have an introduction into that by you just when you were starting FeatureX. It was based on a genetic algorithm, right?

Gil Syswerda: Oh, so were you at that thing?

Sal Daher: I was.

Gil Syswerda: Oh, okay, I was going to ask you about that.

Sal Daher: I understand a little bit about that, surface level, but please explain to the audience what a genetic algorithm is.

Genetic Algotithms: A Lesson from Evolutionary Biology

Gil Syswerda: A genetic algorithm is a way of doing optimization that uses a population of potential solutions. John Holland's inspiration was evolution and nature. If you look out in the world, you have the incredible diversity of plants and animals, all adapt to the ecological niches. There's clearly something really quite powerful going on. You can even harness some of that power by doing breeding and so on. He worked out the underlying mathematics of what must be driving that. A lot of that happens partially at the population level and partially at the genetic level and created an algorithm. You have genetic algorithms, which is really quite powerful.

Gil Syswerda Gets into Hot Water with His Advisor by Being a Know It All

Then he created this thing called the Fundamental Theory of Genetic Algorithms and published a book and became famous for it. His whole career was based on this fundamental theory of genetic algorithms. I'm a student thinking about all this stuff and I start thinking, "Wow, John actually got it wrong," or at least not totally right. His fundamental theory of genetic algorithms wasn't really very fundamental, it was a special case. There was a broader case and if you look at the broader case, chances are the algorithm could be quite a bit more powerful. We had a falling out over it because he didn't like my ideas. It became public in a very unfortunate way.

I'm explaining this because it really had a big impact on me and pretty much every startup I've done since then. That is, we were in a seminar and John was doing his thing and the fundamental theory, blah, blah, blah, so I started raising questions and making John defend whether his thing is general or not. I was kind of an idiot.

[laughter]

Sal Daher: Who's the advisor here and who's the student?

Gil Syswerda: It went on for a couple of minutes. I could see that he was getting really annoyed, so I decided I'm going to stop talking because he's getting mad at me. Suddenly, a voice pops up from the back of the room really loudly and says, "Gil is right." We all turned around and look at this. It was a mathematics professor who had snuck into the back just to see what's going on. He was pretty famous, Quentin Stout.

He starts schooling John Holland on why I was right and he was wrong, but that just sealed my fate. I'm just shrinking down in my seat. Things were never the same between John Holland and I after that. The result of it is that after I graduated from Michigan, I had some fun stories about job searching and so on. If we have a chance at the end, I want to tell you about my interview at Hewlett-Packard and how my third-grade teacher just totally prepared me for a trick question at Hewlett-Packard. It's a fun story.

Sal Daher: Let's remember to.

BBN Labs – Gil Syswerda Wrote Seminal AI Papers that Are Still Frequently Cited

Gil Syswerda: At BBN, I finished that theory and published a paper, Uniform Crossover, and then a follow-on paper, Simulated Crossover. Those papers have had thousands of references in other papers over the years. Still, today, I get notifications once or twice a week that some new paper has been published in reference to these papers. They made an impact in the field. Not only that, I had an algorithm now that worked much, much better than anything before in terms of optimization of certain kinds. At BBN, I was at BBN Labs, a fun place to work.

Sal Daher: BBN was a company that grew out of consulting work of a few professors at MIT, Leo Beranek, and a couple of other guys and consulting during World War II, I don't know, making headsets for when they're flying in the bombers.

Gil Syswerda: A lot of acoustic stuff.

Sal Daher: Acoustic stuff, and it became a very successful research firm and the protocol that is used in the internet protocol, TCP/IP, was developed at BBN. Just to give a context, BBN shows up in other episodes of this because there are a lot of people that I've interviewed who have backgrounds at BBN. Please continue.

Gil Syswerda: One of the things I did at BBN, a job came along for the US Navy. They wanted a scheduling system. I was sent out to talk to them at Point Mugu Naval Air Base in California.

Sal Daher: How did you connect with BBN? You were in Michigan. How did the job come up?

Gil Syswerda: Michigan, I just was mad flurry at the last semester to get up because we had a baby on the way, Carol and I here, we'd been married in the time. Suddenly all my free-living, take classes, I had a little consulting company on the side too to pay the bills. I could have continued that for a long time but the baby really sharpened my focus.

Sal Daher: Random number generator, which is your family.

Gil Syswerda: Jessy was on the way. I decided I needed to finish up school somehow and then go get a job. I ended up getting a master's degree in Computer Engineering with a specialization in microprocessors simply because that was the fastest path I could get for all the classes that I'd taken.

Sal Daher: On your critical path.

Gil Syswerda: Well, I had to do some jumping, I took one class and skipped three prerequisites for that class. Took just the final one and bought all the textbooks for all the other class. It was crazy, crazy busy. Plus, we had a baby right in the middle of all that. Plus, I had a consulting business I just couldn't drop. Then did interviewing. Campus interviews and so on. I did a really fun campus interview at Lawrence Livermore Labs. You can ask me about that one at some point.

Sal Daher: That should be also number two. HP and then Lawrence Livermore Labs.

“What I really wanted was a job where I could use genetic algorithms and the only place on the planet that I could find that did anything of the kind was BBN Labs...”

Gil Syswerda: They had their funny moments in that. What I really wanted was a job where I could use genetic algorithms and the only place on the planet that I could find that did anything of the kind was BBN Labs and that's because the guy there was Dave Davis. I reached out to him and then one thing led to another, I got the interview and finally got a job offer and I decided to make that transition. It was nip and tuck between that and Hewlett-Packard. We did house hunting in Fort Collins because we were getting that close.

Sal Daher: Talk about quality of living.

Gil Syswerda: Exactly. That was the dilemma. I chose BBN because I thought future opportunities for doing interesting things would be better in Boston and working at BBN Labs and I think that was the right choice in the end. Anyway, that's how I ended up at BBN Labs.

Sal Daher: Awesome. You were at BBN.

Gil Syswerda: At BBN, I got heavily involved in the academic field of genetic algorithms, publishing papers. I became associated with the International Society for Genetic Algorithms and became the president of it for several years. We posted conferences of 700, 800 people. It was a big deal. I was also building systems. This Navy problem, I went out there to check it out. I really hit it off of the program manager. I told him, I said, "Okay, I'll take on building a new scheduling system for this lab." It was called a hardware in the loop simulation lab where they did development of the F-14 jet fighter. It's a big expensive lab. Airframes and lots of equipment on.

Sal Daher: Tomcat.

“I tried every way I could think of to get a ride in an F-14 but I never succeeded.”

Gil Syswerda: I tried very hard during that time to try to get a ride in an F-14. I tried every way I could think of to get a ride in an F-14 but I never succeeded. I told him I'd take it on but I was going to use genetic algorithms. He goes, "How are you going to do that exactly?" I say, "I have no idea but I'm going to try." He says, "As long as it works in the end." Then I spent some time thinking about how to do combinatorial optimization on the face of constraints using the genetic algorithm, invented some stuff. We eventually had to file a patent on it, but like an idiot I published it. Then we had to scramble, because it had commercial value, to get a patent filed.

Sal Daher: [laughs] Even brilliant people can do stupid things.

Gil Syswerda Meets Jeff Palmucci and Jeff Herrmann

Gil Syswerda: I just was not thinking commercially at that point. I built a system for the Navy. It was very successful. Then went on and did a bunch of other stuff, computer vision and so on. Lots of BBN Lab type stuff. Cycled back to the scheduling thing because it looked like it might have had commercial potential. I started talking to BBN management about it. We hired a Sloan School graduate student to survey the scheduling market. I had met this guy who gets hired into Labs, Jeff Palmucci hired out of MIT. Jeff and I from that point on worked together for 25 years on various projects.

Sal Daher: Oh, yes. This is interesting. I've seen this before in my interviews.

Gil Syswerda: We also started looking at how to think about a commercial application of this technology. This is Jeff and I working. Met another guy that BBN brought on to look at the marketing aspects of it. That was Jeff Herrmann.

Sal Daher: I know Jeff. We're investors in a couple of companies. He was on the board of SETEM when I invested. It's now XMOS.

Gil Syswerda: Marketing? Okay, great. He's a smart guy.

Sal Daher: Yes.

Gil Syswerda: He was at--

Sal Daher: He's a physicist.

Sal Daher: Studied physics at Harvard or something, if I remember correctly--

Gil Syswerda: Sounds about right.

Sal Daher: Yes.

Gil Syswerda: He wore that math cap for a while. It's a technical marketing background. We made some trips. We got a lead on John Deere, so we made trips out to John Deere with him and so on. Anyway, by the way we realized that the best way to commercialize this is to spin off a commercial company. That company ended up to be Optimax Systems. We had three founders; myself, Jeff Palmucci and Jeff Herrmann. I had to make a big decision about who was going to be the CEO. That's one of the things I want to talk about with startup founders. That's a big decision. I made the decision that I wasn't going to be CEO.

Sal Daher: [laughs]

Gil Syswerda: Now, there's a thing that most startups don't understand, in fact most investors don't understand because they don't ask about this sort of thing.

Sal Daher: Yes.

“Just because you're the CEO doesn't mean you have complete decision authority.”

Gil Syswerda: Just because you're the CEO doesn't mean you have complete decision authority. In fact, you can have the normal corporate documents and then you can have an additional document, sometimes called a shareholder agreement.

Sal Daher: Yes.

Gil Syswerda: A shareholder agreement can cover things like disputes and how you're actually going to make decisions. What the three of us put into place was that any one of us could go off and do our jobs and make decisions up to a certain dollar amount. Past that dollar amount or entering into any contracts or hiring people or firing people, all the big things, we had to get agreement among all three of us. Nobody could just run off and make giant decisions.

The ramifications for not coming to agreement were pretty onerous because we would turn the decision over to somebody else. That was, basically, a couple of board members. That never really happened because we were just too afraid of that, but it really grounded the company.

Sal Daher: Who nominally was the CEO?

Gil Syswerda: Jeff Herrmann.

Sal Daher: Jeff Herrmann?

Gil Syswerda: He had the title.

Sal Daher: It was a weakened role as the CEO.

Gil Syswerda: Wasn't necessarily weakened. He just couldn't arbitrarily make decisions, big ones, without--

Sal Daher: Constrained?

Gil Syswerda: He was constrained.

Sal Daher: Right, this is another thing in Tom Eisenmann's book. Who's going to be CEO?

Gil Syswerda: Yes, well, I learned a lesson from doing Ann Arbor soft works, that first company, because we did not have good decision-making processes at all. Even though I was presumably the CEO of that company, we just don't work together well, that way for instance.

Sal Daher: Right.

Gil Syswerda: I wanted to fix that in this company. This was a much more serious attempt at doing a startup. I would have to leave BBN, which was a good job. I had a mortgage, I had two children at that point. It's just like I'm jumping off a cliff.

Sal Daher: Yes.

Gil Syswerda: It's one of the things. You jump off a cliff, you try to build a parachute on the way down before you splatter.

Sal Daher: [laughs]

Gil Syswerda: It was scary. I was pretty intent on making this thing a success.

Sal Daher: The bad thing is you're jumping off a cliff. The good thing is that your atmosphere is gel, so you're falling slowly. It's a matter of months, it's not in seconds.

Gil Syswerda: Exactly. That whole decision-making process worked, and I would encourage startups to do something of that kind because it makes life a lot easier.

Sal Daher: Constrain the decision-making of the CEO on certain things, give authority in lesser decision-making stuff to the functional people in the company.

Gil Syswerda: We're all doing our jobs, building product, visiting customers, so on and so on.

Sal Daher: How were you funded?

Optimax Systems Was Bootstrapped by the Founders – Had Revenue Right Away

Gil Syswerda: Oh, wow. We spun out of BBN. We each put $5,000, as I recall, into it so we had some operating capital, but we had some real advantages going in. I'm going to introduce a term here, I think of Optimax Systems as a hammer startup. What I mean by that is that if, let's say, I've been managing this, because I know you're going to business school. If you go to business school, and you want to start a business, I can imagine the advice is, "Go find a problem that somebody needs solved, and then build a solution for it."

It makes total common sense. Well, you already said, let's go find a nail, and then build a hammer that hits that nail. Another way of doing it, if you're somebody like me that likes to build hammers; still a hammer that can hit a lot of nails. Then once you have that hammer, go find that right nail.

Sal Daher: You build the tool first and then you look for it to be upscale. Common flaw of engineering founders is that they want to build stuff. They build stuff too early.

Gil Syswerda: Well, you got to build the right thing.

Sal Daher: Well, if you luck into the right thing.

Gil Syswerda: We'll you keep your eyes open. With Optimax Systems, we had already built our scheduling system for the Navy. Now, that was done on a LISP machine and so on. When we decided to commercialize, we realized that at all had to be rewritten in C++. Microsoft Windows 3.1 had just come out so we were going to target that. We talked to the Navy lab that we originally built the system for and asked them if they'd really like to have this whole thing written and run on a PC, as opposed to this expensive LISP machine, so it's not even going to get supported anymore. I went for it.

That was our first contract. It actually went to BBN, they got it signed over to us. Second is, John Deere wanted a scheduling system. We had already been talking to them while we were at BBN. When we span out, first of all, we were really nervous that they were going to go away after we did the startup thing, but there were all for it. We got a contract to build a prototype scheduling system just to see if our stuff worked or not. We got revenue immediately in the door, and then things took off from there.

Sal Daher: Basically, you were bootstrapped with $15,000?

Gil Syswerda: Bootstraped with $15,000.

Sal Daher: Which would be, I don't know, $100,000 today, something like that.

Gil Syswerda: Not even that much, $50,000 or something like. We had revenue coming in, our expenses were light, and things just took off. We didn't raise money until the third year. In the third year, business was really cooking, I got to go back and explain the hammer stuff that we did. In the third year our contracts were chunky. We would get a contract to build a scheduling system for a factory and we won’t to get the actual cash in the door. It was maybe $250,000 until we were done.

Sal Daher: You had working capital requirements?

Raised Money to Cover Working Capital – Michael Mark Invested

Gil Syswerda: Yes, we needed more working capital. We raised $750,000, back then, that ought to be a few 1.5 or something here today. That's when I met Michael Mark, because Michael Mark was an investor. Now, I look back at that, and I think, I wish some of the startups that we talked to as angel investors were like this, because this one was a slam dunk. We had product already in the marketplace, we had customers who were willing to say really nice things about us. We had revenue. We were a going concern. It was a no-brainer to invest in us. A bunch of people raised their hands. Nobody backed out in the end.

Sal Daher: Basically, you were creating a market, you had tools that had massive value, you had a skill that was very rare and, basically, it was an immature business and you guys had what it took to produce things to created value in them and you had no competition. There was not enough capacity, there were not enough people to create things like that. There was massive demand.

Gil Syswerda: Yes, this is where the hammer comes in.

Sal Daher: In a situation like that you can build a hammer first but if it's a competitive situation, and you build a hammer that's why the whole Lean Startup thing came on later because if you have a competitive situation, which you didn't really have in those days.

Gil Syswerda: Well, we did because we were competing against consulting firms and so on. They'll build ad hoc solutions. We're competing against spreadsheet-based stuff. It was all very ad hoc, right. There wasn't a company out there that had a product that could--

Sal Daher: Competed with you.

Gil Syswerda: Yes, and there's good reason for that because it was a very complex situation, right? Every situation we went into had different time models, different resource models, different constraints, different user interface requirements, we had to optimize all that and do it in such a way that we can make money and get in and get out and have a happy customer.

Pitch for Pater Fasse, Patent Attorney at Fish & Richardson

Sal Daher: Okay. This podcast is brought to you by Peter Fasse, patent attorney at Fish and Richardson. The biggest assets of a startup are its team and its intellectual property. Make the most out of the great efforts of your team by getting the best advice on building your intellectual property portfolio. I urge you to check out Peter Fosse patent attorney at Fish and Richardson. Fish and Richardson is the foremost patent law firm and you should not skimp on patent advice. 

Ultimately, you have this astonishing progress. Michael Mark came in as an investor.

Gil Syswerda: Michael Mark came in as an investor and he wasn't on our board, but we gave him board visitation rights.

Sal Daher: By the way, listeners should listen to “The Best Pivots Ever”, which is my interview with Michael Mark. You're not going to find Michael Mark anywhere else in the entire inter webs. Only on my podcast. He's very shy about that kind of stuff but you should listen to that podcast.

Gil Syswerda: To tell the truth, Sal, I have never told anybody any of this stuff. You're the first for--

Sal Daher: I know. This is unique. Anyway, you met Michael Mark, he became an investor, you guys raised them 50 and then?

Gil Syswerda: Yes, Michael’s on our board and I can't say enough good things about Michael Mark. He was the ideal person that have sit in our board, right. We can talk about boards later, and how to have a good board and so on but you need people on your board, who can listen to what you're saying and understand your pain points, and then maybe give you some good advice, right, and Michael was that person. 

I want to go back to the beginning on hammers because it's important. We built a package that was configurable via our own internal language or in that language, you could specify everything a lot of scheduling problem. You could specify the time models, resource models, constraints, user interface components, the whole thing, data in a high-level language, right. Then our system, which is called Optiflex would compile that, and the output would actually be a running scheduling system. Not only that, it would also output the user manual.

[laughs]

“...created a machine for making scheduling systems and user manuals.”

Sal Daher: You created a machine for making scheduling systems and user manuals.

Gil Syswerda: Yes, because describing things like yes, I need to use this user interface component is going to be here. Then you attach as part of the description via the docstrings to it. I got to go eventually in the user manual, right?

Sal Daher: Pretty cool.

Gil Syswerda: It outputed a draft of a user manual. It would take a person to go in there to spend half a day and kind of clean it up a little bit but wow we call this a semi-custom approach because our core technology didn't change. The optimizer didn't change, constraint computation user interface component, those were all C++ libraries, and tested and then those all got put together and configured by this higher-level system into a functioning schedule.

Sal Daher: You were able to sell those as separate pieces of software. 

Gil Syswerda: Well, that's a really important point. That's one of the things we did. Innovative things on the business side here. But one of them was that we did not want to do consulting. It's a different kind of a company and what insisted on was that we go into a company, deploy a system and get out as quickly as possible, with a proviso that they're very happy customers.

“We did not want to try to perpetuate the engagement with a customer by doing consulting...”

We did not want to try to perpetuate the engagement with a customer by doing consulting because that's a very different relationship. Now you're trying to create problems and solve them and just go on and on and on. Our competitors were doing that. We had limited resources. We didn't have that many people and we wanted more customers. We wanted to turn these systems around as quickly as possible. That drove the technology and let us do this thing.

Coming into Optimax, we didn't know that we were going to be doing factory scheduling. We knew that we had a hammer, we knew what we wanted to build, and we knew the capabilities, and a lot of the C++ libraries were already half built. The whole 4GL thing wasn't quite yet but we were on our way. This is a safer way of doing a startup because you talk about pivots.

Sal Daher: [laughs] Yes.

Gil Syswerda: Pivot-happy companies out there, right? I'll just say it. Pivots are a bad thing.

Sal Daher: No. I mean yes. The alternative is for the company to go belly up.

Gil Syswerda: Yes. Pivots are a necessary thing but you don't want them to be necessary, right?

Sal Daher: No.

Gil Syswerda: It's really not a badge of honor to say I've done five pivots so far.

Optimax Systems Followed Its Business Plan; Never Pivoted 

Sal Daher: No, but according to Michael Mark, of the 200 companies that he'd invested in at that point, how many companies went according to the business plan that didn't have to pivot? None. Oh one, Progress [Software].

Gil Syswerda: Yes. Actually, in a Walnut meeting, he mentioned Optimax systems.

Sal Daher: Oh, okay. Maybe there were two out of 200, okay.

Gil Syswerda: Yes, because he says, "There's only one--" He didn't quite say not pivots. He says, "There's only one company in my entire experience that actually executed their business plan." Perhaps I’d go up to the right. At the beginning, you had to sell the investors, right? He says, "And that was Gil's company," and I'm going, "What? Really?"

[laughter]

Because who reads their business plan after he got started? Apparently, Michael was. We clocked along and actually, hit our targets but we're going to back up again to pivots, right? If you have a hammer company, you have a tool that can hit a lot of nails, right? In the beginning of the company, go hit a bunch and figure out what nail is the right ones. I'm just going to list the things that we did at the beginning of the company. A couple were done before we left BBN.

We looked at routing of conduits in Navy ships because a Navy ship is a 3D mass of pipes and wires, right?

Sal Daher: Yes.

Gil Syswerda: When they want want to put something new in there, there's got to be rerouted and we looked at Navy fleet scheduling and we built a prototype to schedule the entire Pacific Fleet. Then went and showed at CINCPAC fleet headquarters in Hawaii.

Sal Daher: You got to fly to Hawaii.

Shell Oil’s Tank Farm Broke Opitmax’s Optimization Algorithm

Gil Syswerda: Yes, we got to fly to Hawaii. Saw a volcano. We did a John Deere assembly line prototype. We looked very closely at oil pipeline scheduling for Shell Oil. Roughly 200,000 miles of pipeline in United States that ship the liquid products. Things like kerosene, jet fuel, gasoline and so on. It's complicated because there's different batches of product going down these pipelines and there's demands and so on, so there's way stations along away. Tank farms may siphon some of it off and send it in different places and so on, so it turns into a pretty big scheduling problem.

It was cool. We decided not to do it because it had a technical aspect. This is where details matter and why we looked at this thing so carefully. That problem had a resource, a tank where product could go in and go out at the same time on a different pipeline. One pipeline is putting the product in, the other pipeline is pulling it out. That broke our optimizer.

[laughter] [crosstalk]

Sal Daher: We got a leak of the system. [laughs]

Gil Syswerda: We can handle that particular resource and we didn’t want to try so we walked away. We looked at packing shipping containers. If you're a big manufacturer, you're a global company, you have a factory somewhere. What happens, a lot of parks shop at that factory, right? They get shipped over and they get put on a truck and a truck shows up. It's a loading dock. You open the back and there's a whole bunch of stuff in there. They want the stuff that's at the back of the truck to be what's needed on the assembly line that day.

It doesn't do them much good if the thing that they need is in the front of the truck or in the middle of the truck because now they got to unpack the truck, and they got to figure out where to put all that inventory. They just wanted to streamline the whole thing and doing the global way. That's a big scheduling problem. Oh, by the way, they wanted us to tightly pack the shipping container as well, which is a bin packing problem and we could do that as well. Anyway, one thing we did look at too was patient scheduling at hospitals. That was to the Mayo Clinic. We had some deep discussions with Mayo Clinic.

They came to us initially, they said, "Just treat us as a factory. Sick patients come in; stuff happens to them while patients go out." That looked like an interesting area to us as well. 

The point I'm making is, we hit all these nails, built prototypes, or at least consider them carefully and then chose to do two of them. We chose factory scheduling, assembly line of heavy equipment like tractors and cars and things like that. We actually took on Mayo's problem, to do patient scheduling.

We chose Mayo not because we wanted to do patient scheduling so much, it's that Mayo's problem required certain capabilities that we knew we would need in the future and they were willing to pay us to develop them in conjunction with solving their problem.

Sal Daher: Ha. It was strategically important.

Built a Scheduling System for the Mayo Clinic, Made the Mistake of Thinking MGH Would Buy it Too

Gil Syswerda: Yes. It looked like it might be a good business area because we thought, okay, once we've done Mayo, maybe Mass General will like the system too.

[laughter]

We had never done any research on it at all, right? It turns out that's not true.

Sal Daher: Yes, not at all. They're going to build their own internally.

Gil Syswerda: Yes, it's like we don't want Mayo's system. We want our system. There's a really big [crosstalk]

Sal Daher: Talk about not [laughter] feeling.

That's the engineer thinking. Let the engineer-- Never let the engineers do the marketing.

Gil Syswerda: We weren't [laughter] completely blind because we actually partnered with Hewlett Packard because they had a team that always wanted to do electronic medical records. We thought we could jointly go in; medical records and scheduling, they go together. Wow, they got the cold shoulder too, right? We got a call from them eventually, says, you guys, we're getting out of this market. We can't sell into the medical space.

Sal Daher: Let's skip to the ultimate result of Optimax.

“It was a very rapid four-year run to get an acquisition offer from i2 Technologies. We beat them in every single sales situation.”

Gil Syswerda: It was a very rapid four-year run to get an acquisition offer from i2 Technologies. We beat them in every single sales situation. We could get our toe in the door, before a contract was signed, we took it away from them. That was true for them. That was true with everybody. We never, ever lost in the marketplace because we had a very compelling [crosstalk]

Sal Daher: You were technologically dominant.

Gil Syswerda: We are technologically dominant; we are also business dominant. Some of the things we did-

Sal Daher: You could execute.

Gil Syswerda: -we offered fixed price. We offered to deploy on our own hardware on a PC. It had to be a high-end PC. No data center stuff and so on. We had a really powerful user interface for factory people. They felt like they're in control of this thing, even though 99.9% of the time is fully automatic. It was just a very compelling solution.

Sal Daher: Excellent.

Gil Syswerda: We were unstoppable. Nobody could touch us. The only one that we lost in the very end, because they found out that we were getting acquired, was Harley Davidson.

Sal Daher: What did that do for you when you had that exit?

Gil Syswerda: There was a process that we went through to do the exit. We only had one buyer. We could talk about-- I know we're going to run out of time here but there's things you can do. We hired an investment banker to find us another buyer and stuff, or more than one to work that process. Then in the end, we got an offer we just couldn't refuse. 

Optimax’s Revenues Had a Direct Positive Impact on i2Technolgies Earnings

Got acquired by i2. i2 at the time was a public company, about 6000 employees, roughly. We were about 35. Another thing by the way, that was really important to the acquisition, and this is a point I make to startups I advise, is that the purchase of our company was accretive to earnings.

The reason it was is because our revenue came from license revenue, not from consulting revenue. When we make a sale, you could recognize that revenue immediately even if you weren't necessarily paid, they just go into accounts receivable. That pulled the revenue forward. That could be a make-or-break kind of a thing.

Sal Daher: They were buying EBITDA.

Gil Syswerda: When they announced it, they could just tell their investors, the public markets, that this is accretive earnings.

Sal Daher: Their stock valuation would go up.

Gil Syswerda: Yes, it did. [laughs] [crosstalk]

Sal Daher: Instead of the [laughter] usual acquisition where they acquire and then things go down and value of the company goes down because they don't know if they're going to be able to mesh the two cultures and all that stuff. In this case, they were buying revenue, so to speak.

Gil Syswerda: They were buying a marketplace too because we were in the automotive sector and they really wanted to be in that space. They were buying technology. They did not have detail production scheduling, so we fit a pretty big hole in their product suite. We had really good people. There was a lot of reasons to buy us. It was really quite a successful merger.

Sal Daher: What did that do for you personally? Then were you at a point where you were independently wealthy at that time?

Gil Syswerda: I owned i2 stock, so I could [crosstalk]

Sal Daher: You were paid in stock.

Gil Syswerda: Yes. Which you could then sell after a while. It was a decent size, piece of change. In the conversation with the CEO, founder of i2, he wanted to tie us down somehow. His name was Sanjiv Sidhu. I said, "Look, we actually like our jobs and we like to work. Instead of trying to figure out how to tie us down, why don't you try to figure out how to make us happy on a day-to-day basis with the work that we're doing?" He totally bought it. We had no restrictions whatsoever. We could have walked away the next day after we signed the docs but we didn't. We stuck around.

We first of all, worked really hard to make the merger successful because we figured we owed that to them. Then I was asked to become the CTO of i2. I declined because I didn't want to move to Dallas. I thought I really needed to be in an executive role to be really effective but I agreed to become the acting CTO and to help them find a real CTO, which we never did. I realized that I have some capabilities that a lot of people don't have because I could not find my replacement. That is, I'm really good at technology and have become pretty good at business.

That's what you really want in your CTO. You want both and good luck finding those people. There's not too many of those around. The ones that are, are probably doing either their own startup or they're heavily protected by their employers. It's hard to find people like that. Anyway, I did that for several years and it was like grabbing a tiger by the tail. This is in the .com run-up. i2 was right in the middle of it. The stock was going up. It was a lot of fun there for a while.

What Gil Syswerda Did After Leaving i2 – Travel the World + Race Cars

Then things got crazy busy and I didn't really have to be working that hard. I renegotiated my employment agreement with i2 where I could start working on my own stuff and not have any IP conflicts and reduce my hours. I was put on basically, a retainer, against hours and or something. Then I did some really fun stuff. Traveled the world with my family. Did some really interesting trips. Another thing I did is I became a really serious race car driver.

Sal Daher: This sounds a little bit like Russ Wilcox after they sold the digital ink company that they had.

Gil Syswerda: E Ink?

Sal Daher: E Ink, yes. E Ink. He took a whole year off and they went around the world, traveled with his family, and so forth. You became a race car driver.

The Big Lesson for Startups from Being a Race Car Driver 

Gil Syswerda: I traveled and became a race car driver and started working on some different technology, which probably, you don't have time to talk about that. I do want to talk about race car driving in a second. There's a big lesson in race car driving that applies to startups. I'm got to make a point and it's probably the far most important point I'm going to make. If you don't take anything else away, take this away.

Sal Daher: Exactly. Does it help you with your career race car driving or with startups?

Gil Syswerda: Both. There's an analogy. They both apply. I learned to become a race car driver knowing nothing. I'd never been in a race car. It was a two-year process of learning to drive the car, building up my skills, learning how to deal with traffic on a race track, and so on, to the point, where I could win races. Now there's a lot of things that go into winning a race but here's a critical thing that you need to do as a race car driver. That is, you need to look at where you want your car to go.

If you're going a hundred miles an hour, let's say, you're moving down the road at the rate of about two tennis courts per second. You want to look probably further than that, 300, 400 feet down the road. It varies, depending on the conditions. Your car tends to go where you look. If you're looking at the bumper of the car in front of you--

Sal Daher: You going to go in the bumper of the car in front of you. [laughs]

Gil Syswerda: Yes, you are. One of two things is going to happen. You're either going to hit that car, right?

Sal Daher: That's right.

Gil Syswerda: Or you're going to do exactly what that car does. If they screw up, then you're going to screw up. Also, another thing you absolutely cannot do is, and this happens when you're first learning to drive a race car, is, think about if you screw this turn-up and you spin out, where's your car going to go? Because then you end up looking where your car's going to go off the track.

Sal Daher: Then your car's going to end [crosstalk]

Gil Syswerda: It's just absolutely critical.

Sal Daher: Gil, you must have missed drivers' ed, because in drivers ed, they teach you how to stay in the lane because you don't look at the lane markers, you look at forward, you look far up.

Gil Syswerda: Yes, you look ahead of the car in front of you.

Sal Daher: Ahead of the car, yes. [laughs].

Gil Syswerda: You try that when you drive at that speed in a race car. Magic things happen if you do that. If you're coming into a turn, there's a car in front of you, you want to pass that car, you're not looking at the track right there. You're not looking at that car. You're looking down around the corner of the curve because that's where you want your car to be. A lot of things just start working themselves out. Your background process starts figuring out how you get to deal with this car in front of you and how you going to possibly pass it going into the turn and so on.

You can get to be a pretty good race car driver without doing this look ahead thing. You cannot win races. You will just likely be a fifth-place driver if you don't, so looking ahead is critical. 

Looking ahead is critical when you're doing a startup as well. When you're in the middle of a startup, and by the way, for all those who are in a startup right now, startups are really designed to make you feel stupid and inadequate in the moment. No matter how much success you've had, it sucks.

Sal Daher: You're doing stuff that's totally new and you're going to feel inept.

Gil Syswerda: It's new, your to-do list has hundreds of items on it, you're trying to figure out what's important. There's all these issues hitting you left and right. We're not quite following a straight-line narrative through all the startups I've done, but I could tell a story where it all sounds logical and I did this and I did that and I had the success, I had that success, and so on. It's not like that when you're in the middle of a startup. I don't want to forget that point. I was going to make that point earlier. It's like, if you're in a startup and you're feeling inadequate, it's normal. It's normal.

Sal Daher: You need to look forward. You look ahead.

Gil Syswerda’s Management Workflow at Optimax Systems

Gil Syswerda: Rewind back to Optimax Systems. The way we ran that company and the way I personally did it is, every week, on Sunday mornings, for about two hours, I would mentally review all aspects of the business, but not the aspects of like, I need to get a job offer out or somebody's unhappy or [crosstalk]

Sal Daher: Not the checklist.

Gil Syswerda: It was just a mental review, where do I want this company to be by the end of the week, by the end of the month and, and so on. Look forward. You have to do that at least once a week. Then on Monday mornings, we would have what was called a staff meeting. It was three founders for sure, then we'd bring other people in, then we had some more senior people later. Every week, this is a four-hour meeting, every week we would review all aspects of the business. Every last little thing.

All the development projects that we had going on, product enhancements, every customer that is in process or in the queue, our whole sales pipeline, we would look at our finances and when we expect money, took a cash view of the business. There was no part of the business that we didn't review every single week. All projected forward so that we can make plans for what's coming up. Then after that, this is on the next day, we'd have a resource planning meeting, and this is usually Jeff and I, where we would plan what everybody was going to be doing for the next week. We ran the whole company on a weekly cycle but that Monday morning meeting, was looking forward.

Sal Daher: That set the tone for the whole week, that Monday morning meeting?

Gil Syswerda: It set the tone for the week, for the month, and so on. We would look forward as far as we reasonably could because we wanted to see what roadblocks we had up ahead, and so on. It kept everybody's mind focused on where the business is going. It just became part of our culture. We're not looking at, "My God, how are we ever going to get this little thing done or how are we going to-- We have an open hire that we haven't done for a while," and so on.

Sal Daher: Yes. Which is typical in a startup because you're under-resourced and you're always looking right just in front of you.

Gil Syswerda: You just get blasted by stuff all the time. It was very hard to not have a really myopic view. Basically, you're looking at a road right in front of you, you're not looking at the turn up ahead.

Sal Daher: How much did your board influence that?

Gil Syswerda Believes in Putting Boards to Work

Gil Syswerda: I would flip it around. My relationships with boards has been that, put them to work. That process that we went through every Monday, when the board meeting rolled around, guess what? We're going to hit them with all that stuff.

[laughter]

There was no, let's pretend that they're actually investors, which they were, and paint them a happy picture. We would hit them with every last detail that we were struggling with and what was coming up and so on.

Sal Daher: There was a roll-up-your-sleeves board?

Gil Syswerda: Roll up your sleeves board. If you don't like that sort of thing or you're not capable of understanding that level of detail and so on, then you shouldn't be on this. One of the boards might [crosstalk]

Sal Daher: The board I'm on is like that. The board meeting is coming up on the 17th. [laughs] Everything is going to go on to the board meeting, because we've been grappling with that all along.

Gil Syswerda: Grappling with?

Sal Daher: With all the stuff that we're going to be talking about in the board meeting.

Gil Syswerda: Yes. The board shouldn't be caught by surprise by a lot of stuff. There needs to be a lot of communication, right?

Sal Daher: Yes. It helps that it's a three-person board, plus the founder.

Gil Syswerda: Got you. I've always taken the attitude is that the board is there for the company, they need to be working.

Sal Daher: Absolutely. Yes. For founders, this is really important. Don't assume your board is just something you have to manage. That it's like a bunch of wild beasts you got to throw meat to. Your board is a tool that you got to use. What Gil is saying here is really valuable. Don't waste the tremendous resource that the board can be.

Gil Syswerda: Yes, and you're going to get a conflict because sometimes investors end up on your board. I think it's a mistake to continue to treat them as an investor because investors, you paint a pretty picture, right? You can fool investors to maybe a couple of rounds of funding before things get really real and they realize you have no customers, right? You could do the same thing with your board.

Sal Daher: You could be in a board of some founder and fool the investors.

Gil Syswerda: I have been some bad board meetings, not my company.

Sal Daher: No. A company you are on the board of, I'm saying.

Gil Syswerda: I'm sorry, what?

Sal Daher: I don't want to mention the name but a company that you are the board of where the investor went more than a few rounds.

Gil Syswerda: Yes.

Sal Daher: The founder.

Gil Syswerda: Yes. I know a company. I don't want to mention names either and I really fought pretty hard in that board because the other board members that they had just wanted to be happy. They perpetuated the happy view of the world. I tried to explain to them, "We're listening to a fantasy here. This is not a sales call. Just drill down and figure out some of these metrics of where this company is going." It's a very little interest on anybody's part to do that.

Sal Daher: Yes. Please continue. Basically, looking ahead and--

Gil Syswerda: Yes, it's a very important point. Structure your companies that way, your company culture, so that at least once a week you are as a team, looking ahead. You're not just-- You're all nose to the grindstone and trying to figure out what we're on. If you do that, by the way, you don't really need a board, because you're basically operating as your own board.

Sal Daher: This is Ed Roberts, the professor emeritus now, of entrepreneurship at MIT. He says that's what a board is for, is to make sure that the founders are looking beyond their shoes.

Gil Syswerda: I know, but it's really hard as a board member, to make founders do that.

Sal Daher: To get them to look up.

Gil Syswerda: I've tried. You can't just jump in once a month or once every three months or whatever, and what? Redirect the organization to look down the road of it.

Sal Daher: Right.

Gil Syswerda: That's a heavy lift. It doesn't really work. The founders themselves should be doing it and then rope the board into doing it with them.

Sal Daher: Okay, so we have your first amateur effort at founding a company, then your big success when you found the company with Jeff Herrmann.

Gil Syswerda: Jeff Palmucci.

Sal Daher: Yes. Next, you're building your own stuff, you were at IT2 or whatever the company was called.

Gil Syswerda: Yes, i2, sorry. Did there for several years, very intensely. It was a lot of fun. I really liked working with the team there. Backed off, did the whole race car driving thing, travel, but also had to decide what I was going to do next. One of them, by the way, is I could've become a professional race car driver.

[laughter]

I am not kidding you. I got really good at it. Me and another guy that I was working with, we were starting to win a lot of races.

Sal Daher: I think you've got a little bit of AGI inside you. [laughs] Looking across the first instance of AGI.

Gil Syswerda: Artificial General Intelligence?

Sal Daher: Yes. [laughs]. He's not a human. You could have become a race car driver. Really great sports people, they can play any sport.

Gil Syswerda: I didn't know I had it in me to be a race car driver. That was a complete surprise. In the series that we were in, I would race on a weekend, there'd be professionals. I would be racing against professional drivers because they would come into our series when they weren't out there because they didn't have a race that weekend. To keep themselves fresh, they would race against us. Then the next weekend, I'd be watching a race on TV. I'd tell my wife, I said, "See that guy, I beat him last weekend."

[laughter]

I realized he was like, I could be at the [laughter] of that race. Anyway, I came to my senses.

Sal Daher: In AI it's rare that you get incinerated inside a car from a crash.

“...instead of becoming a professional race car driver, I decided to go back to my roots and build another hammer. Again, it's based on genetic algorithms...”

Gil Syswerda: Yes. It doesn't happen very often. It's not without danger, but it's not as dangerous as a lot of people think. What I decided to do instead of becoming a professional race car driver, I decided to go back to my roots and build another hammer. Again, it's based on genetic algorithms, but now, what I wanted to do was build a rule induction system. Those rules would be such that you could give it a data set and it would have rules that apply to that data set and basically, learn how to describe regularities in that data set.

Sal Daher: Wait a second. When is it that you joined Walnut? Were you still a race car driver when you joined? Were you in your race car driving?

Gil Syswerda: I should have looked that up. I asked Michael Mark what his memory was of when I joined Walnut. I've been a member for a long time but I've been in and out. I've always paid my dues, so I've never not being a member, right?

Sal Daher: Now the dues are $400 a year.

Gil Syswerda: Yes, I know.

Sal Daher: It's not much.

Gil Syswerda: It's not much but I've never been asked to leave either because I'm not attending meetings anymore. When I'm between startups, I attend meetings.

Sal Daher: It's a New England investment club, $400 bucks. There's some members of grumble. [laughs]

Gil Syswerda: I think I might be up to $500 now. It pays for little wonderful dinners that we have. When I'm in the middle of a startup, I just don't have time.

Sal Daher: I remember you disappeared. You've had two cycles of disappearing the time that I've been in Walnut. Please continue. The genetic rule induction, because this beginning to sound like the stuff that presentations [crosstalk] Precursor. Okay.

Gil Syswerda: I started making some progress. I pulled Jeff Palmucci in. We banged our heads against this problem for quite some time, six months maybe, trying to get it to go. It was probably the hardest intellectual thing I had done up to that point. Get this thing to work. The precursor was actually going back to John Holland, his classifier systems. He had invented the idea of having a role induction system but he never got it to work and 15 years of grad students didn't get it to work either. Now we're trying to get it to work.

It turned out we had to change a lot of stuff and we invented some fundamental things along the way and eventually ended up with a system called ECS, Explicit Classification System. It could find patterns in data that no other machine learning system could find for certain classes of problem, useful classes of problem. It just ran circles around the state of the art in machine learning. That's cool. Now we have a hammer, find patterns in data. Now we got need to find a nail, right? We formed a company, Machine Insight.

“Machine Insight was to correct a really big tactical error that we made with Optimax Systems.”

Machine Insight was to correct a really big tactical error that we made with Optimax Systems. This was pointed out to me, by the way, by Dave Blundin from Link Ventures. I don't know if you know him or not. He'd probably be an interesting person to interview.

Sal Daher: Sure. Love to talk to him.

Gil Syswerda: Dave pointed out to me once, he says, "What you should have done with Optimax because you had this package Optiflex that could--" The hammer, it hit lots of nails. "What you should have done is you should put the hammer in a different company, an IP holding company. Then build Optimax Systems that has licensed rights to use that hammer for whatever you targeted, detail production schedule."

Sal Daher: Create a platform, then spin off different verticals from that platform?

Gil Syswerda: Right, then when you sell Optimax Systems, that's fine. i2 can use it, just-- I mean they're not restricted at all to apply to their business. Right?

Sal Daher: Right.

Gil Syswerda: That hammer could have hit a lot of other things. We could have done chip layout with that thing, vehicle reides, all kinds of stuff, right?

Sal Daher: Yes.

Gil Syswerda: We sold the whole thing lock, stock, and barrel, all future potential market uses for that technology on the sale, which, like, "Oops." [crosstalk]

Sal Daher: Yes. You didn't build a hammer, you built a platform, a hammering platform.

Gil Syswerda: Right. Not going to make that mistake this time, right??

Sal Daher: Right.

Gil Syswerda: We build this hammer, ECS. It resides now in this new company, Machine Insight. Then we went shopping around. We built some demos and so on. I remember one day-- The night before come back from John Deere, because we visited a lot of our old customers to see what their needs were. John Deere had a really interesting problem, analyzing warranty information to identify underlying faults in their equipment and so on.

It was ten o'clock in the morning and I get a call. We had offices by then in Harvard Square for Machine Insight. I get a call from Dave Blundin, who knew what we were doing. He says, "Gil, I have a whole bunch of financial people in our office right now. How would you like to come and give a demo?" I said, "Sure. When would you like to do it?" He says, "Right now."

Sal Daher: [laughs]

Gil Syswerda: "Whoa." I was like, I was locked and loaded because I just got back from John Deere. Right?

Sal Daher: Yes.

Gil Syswerda: I drove from Harvard Square to out to Wakefield where their offices are, and had a room full of financial people, all these grey suits. I start giving the pitch, it was a technology pitch. I got like 2/3 of the way through that pitch and I realized that half the people aren't even listening to me anymore. They're talking to each other.

Sal Daher: [laughs]

“I didn't even have to finish the presentation and they're already sold.”

Gil Syswerda: What I realized what they're talking about, is how they're going to use this technology to make a killing in the market. I didn't even have to finish the presentation and they're already sold. We've got just an amazingly enthusiastic--

Sal Daher: Roughly what year is this?

Gil Syswerda: I'm terrible at year, so let's see. We started Percipio after that, so Percipio was started 2004. Maybe this is now 2003, I guess.

Sal Daher: 2003? Okay.

Gil Syswerda: We shopped-- It was just the technology. It wasn't very big hammer. It wasn't a very well-developed hammer, it's all together with bailing wire and chewing gum. It could do machine learning stuff but we hadn't applied it to a lot of things. We hadn't applied it to financial data at all, but we've shopped it around, we met with a lot of financial firms. Goldman Sachs, Wellington, Putnam, also a bunch of the-- It really came down to one of three avenues. We can license the technology to a firm, and we weren’t really ready for that. We could partner with a firm, which meant that we be locked into just one company now, doing stuff for them, or we just start our own financial firm.

Sal Daher: [chuckles]

Gil Syswerda: After thinking about it? Jeff and I, just sitting there--

Sal Daher: [laughs] Why not? If you could be race car driver you could be a fund manager.

Gil Syswerda: Maybe we should just start a hedge fund, right?

Sal Daher: Yes.

"Do we even know what a hedge fund does?"

Gil Syswerda: We had a conversation about it. He says-- It's like, "Do we even know what a hedge fund does?" Jeff says, "I think they're allowed to go short as opposed to just long." Whatever that meant. I said, "What's the hedging part of a hedge fund?" He said, "I have no idea." I actually, honestly got-- I got on Amazon and I typed in hedge funds for dummies. [crosstalk]

[laughter]

We just didn't know, right?

Sal Daher: Yes. The point is not where you start. The point is where you end up.

Gil Syswerda: Yes, exactly. I just found some introductory books because I needed to learn the terminology and so. It didn't take us very long to get up to speed on what a hedge fund was, but there's also books on financial fundamentals of investing and so on, and options trading. We bought some of those books, and oh, my God. I could just shoot myself trying to read some of that stuff.

What we also did was, we went looking for advisors and we found really great advisors. Dave Blundin wanted to join. He's both a technology guy and had a financial background. Got Mark Cassidy, he's the CEO of a billion-dollar financial firm downtown, at the time. He's retired off of that now. Rock Hillenbrand, ex-partner at Goldman Sachs, and so on and so on, a bunch of people. The deal was-- By the way, they all wanted to be on the board. The deal was, you put some money into the company, [crosstalk] investment, so you can be on the on the board.

Sal Daher: It's like a non-profit board, you all have to put money in.

Gil Syswerda: You'd have to buy your way into our board. By the way, we're going to put you to work if you do that. Then we launched Percipio Capital Management. Through our contacts, we found a guy who had run big funds and so on, had deep financial experience. We made him the CEO, same kind of a deal, brought on as a partner and decision-making process and all that.

Then Jeff and I went to work, trying to figure out how to build a product for the financial space. We decided to tackle both the financial markets and commodities, trading front-month contract, and ran into a huge litany of problems. The biggest one we ran into is an issue called data snooping, which we just were unbelievably unaware of. Machine learning people, they talk about it now, but back then, it was just like, nobody was aware.

Data snooping problem is, if you have a big, noisy dataset and you go into it and you're looking for patterns, if you look long enough, you're going to find something because the more you look, the more likely it is you're slicing and dicing this data [crosstalk].

It's a little bit like flipping a coin. If you flip a coin 20 times, what's the chances you're going to get all heads? The chances are really low. It turns out to be about one in a million. Now you flip a coin a million times and you go looking for 20 heads in a row in that whole stream of flips, your chances of finding it are about 50%. If you go and looking and you find suddenly, 20 heads in a row in your data, and you think, "Wow. That is really unlikely. This is really significant." Ordinary statistical tests will show us significant, right? But it's not. It's spurious.

Sal Daher: It's data mining.

Gil Syswerda: Yes. Data mining is used in bunch of different ways. I don't really like that term because it confuses people. Snooping is a real issue. Over training in machine learning is an issue too, but snooping is even worse. It took us a good part of a year to figure out how to not snoop. There's mathematical stuff behind it and all that stuff doesn't work for the kind of things that we were doing.

Sal Daher: You had to do falsifiable experiments. You had to be like a scientist stating a hypothesis and then going out there and testing your hypothesis against the data and falsifying it, or not falsifying it.

Gil Syswerda: Yes, it's a comparison-type problem that you run into, so that doesn't work. It's like, let's say you're looking for a place to put your money and you get a list of all the mutual funds out there or the ETFs or whatever. You look at their performance over time. You look at the performance from the last month, the last year or whatever, so you compare 10,000 of these against each other and pick the top one.

[music]

More Gil Syswerda to Come in a Future Episode

Sal Daher: Hey, this is Sal Daher. We've got to stop right here. We've covered a lot of material in this episode. We're going to have a separate episode, which is going to be mostly about artificial intelligence and the future. I hope you've enjoyed this first instalment of our conversation with Gil Syswerda? There's more to come. Thanks for listening. This is Sal Daher.

[music]

I'm glad you were able to join us. Our engineer is Raul Rosa. Our theme was composed by John McKusick. Our graphic design is by Katharine Woodward-Maynard. Our host is coached by Grace Daher.