March 22, 2018

Starting a Business in Silicon Valley

A Personal Story

(This is the text of an essay I read in a presentation at African Leadership University in Mauritius. Please see the video of that presentation here. For the other lectures from that trip, please see this playlist.)

When it comes to starting a business, Silicon Valley is kind of a special place. You can start a successful business anywhere in the world, but no other place has the home of Google, Apple, Facebook, Tesla, and Netflix. Startup accelerators like Y Combinator and 500 Startups seek out new and perhaps inexperienced talent and give them the money and guidance they need to serve billion dollar markets. Angel investors rich on the Silicon Valley gold rush try their hand at investment, writing $50,000 checks in the coffee shops of Palo Alto - the birthplace of the personal computer. Stanford University pushes out some of the world's most talented minds and nearby Sand Hill Road feeds them cash from billion dollar Venture Capital funds.

Silicon Valley is the place where Mark Zuckerberg - a college drop out - was given $500,000 to continue development on a website that made no money. Zuckerberg, who is the same age I am, is now worth $70 billion.

I grew up just 45 minutes from Silicon Valley in a small town in the Santa Cruz Mountains called Ben Lomond. My father was an Electrical Engineer from Lockheed Missiles and Space, whose first job out of college was developing guidance systems for rockets at Vandenberg Air Force Base. When I was two years old he was transferred to their Sunnyvale facility, and one of my earliest memories was my third birthday at a hotel in the mountain town I would call my home for almost 20 years.

Ben Lomond is a town filled with 1000 year old redwood trees that tower above you at 100 meters tall. The trees draw in so much fog and rain from the ocean that the valley I lived in is considered a temperate rainforest. When I was very young my parents separated, and I spent my childhood safe at home while my mom worked, I remember always the rainfall through the picture window. The nearby university focuses on environmental protection, and I was raised with an appreciation for our Earth and all that it provides.

When I was four years old, I told my parents I wanted to be an inventor. At age 5 I took the film winding motor from an old camera and added it to a plastic toy car - perhaps my first invention. As I got a little older my engineer father began to insist "inventor" isn't a job. And so at age 11 when I watched a show on robotics on The Learning Channel, I knew at that moment I wanted to run my own robotics company. I have been hell bent on it ever since.

When the search for college came around, I found a school close to home that had a great robotics program - Santa Clara University - and finally landed myself in Silicon Valley. I didn't know it at the time, but I'd be living in the area for 15 more years and counting.

In college I started to notice other businesses people had founded where they were able to work on their dreams. Sparkfun Electronics sold open source gadgets for hobbyists. I was amazed when, after 8 or so years in business, they shared a plot of their year over year revenue figures since their founding. They had doubled in size every year until they were doing $30m in sales annually. I've always been smart and confident in myself, and so I felt strongly that if others could do this well, so could I.

I would eventually drop out of college - I was more interested in staying home and building robots than going to class - and worked for 7 years doing Mechanical Design and manufacturing at a small engineering shop in the area.

It was during my time in that position that I really started to experiment with engineering at home on my own time. I was learning some wonderful skills at my day job and I wanted to apply them to something interesting. I started to learn electronics design at home. I spent evenings reading tutorials online and using free software to design LED lighting for my office long before you could buy LED lighting at the corner store. I learned how to read manufacturer's datasheets and turn specifications and diagrams in to functional designs. I ordered circuit boards from china and learned surface mount soldering at home.

All the while, my goal was to learn every facet of robotics. I did mechanical design at work, electronics at home, and started to dabble in more advanced software as well.

Eventually I started to put it all together. After several years doing design and manufacturing at my day job, I started to work on software projects there too. I designed a tablet computer from scratch and integrated electrical components from different vendors to make it a reality. I put together a touch panel over an LCD screen, found a single board computer I could program, designed a case for the whole system and wrote all the software for it. At home I started exploring the design of custom brushless linear motors - devices with one moving part that turn electrical energy in to linear motion. I designed the electronics, wrote software, and machined the mechanicals. I even made the electric motor winder out of an old drill motor and a beer bottle.

After 7 years at my day job job, I was itching for something new. Finally I found a famous speech from Steve Jobs - his Stanford Commencement Speech - that said roughly: If you wake up every day and you don't like what you're going to do, then go out and change it. The world is not run by special people somehow more capable than most, but regular people no smarter than you or I who found themselves at the right place and the right time. If you're not happy with your life, then don't keep doing it. Find your passion and follow it.

I knew that the job I was working at - Mechanical Engineer at an industrial tool supplier - was not my passion. I wake up every day and I see a world that could be made better with robots. I see people going to work doing things they don't want to do, and I see how we could make robots to help them.

So I quit. I had no plan and no money. It was probably a bad idea, but somehow it worked out. My employer didn't want to lose me, and we worked out a deal where I would agree to stay full time for a few more months, and then stay on one day a week for another year while receiving my full salary. Honestly I feel guilty for that - it doesn't seem like a fair deal to my boss, but it was his company and it seems he wanted me. I'm still indebted both figuratively and literally for all the help he gave me while we worked together.

But I got my chance. I wanted to form a company around the linear motors I was developing. I didn't know anything about how to run a business. People asked me what my business plan was and I said "make hardware and sell it". What I would make exactly or who I would sell it to or why they would buy it from me were questions I had no answer to and no strong motivation to answer.

I spent months designing new hardware - winding motors, writing software simulations, designing new PCBs, 3D printing housings, machining aluminum, and making calibration fixtures. I went to a trade show and set up a booth and people loved what I was trying to build.

But after pouring months in to the project, I was scarcely closer to production than I was when I began. I had a fabulous prototype, but not a product. I didn't even understand the distinction at the time.

But this is Silicon Valley - a place flooded with investment in all kinds of crazy ideas. Mobile apps were all the rage, but there exists a small faction of engineers and start ups in the area that are focused on making hardware. I regularly took my prototypes to the hardware startup meetups and was looking for any investment I could find. I made friends who had gone through Stanford Business school who tried to tell me I was asking for too much money and lacked a solid business plan. I wanted $500k investment and I was the only employee!

Eventually I would learn the hard way that the money doesn't come that easy. I had been reading Techcrunch - a valley blog focused on the goings on of investment in the tech world - and all the stories of success let me believe that anyone with an idea could find investment. But my product was complex. I had no market. And I had no team.

I was working with a friend for a little while on the project - a seasoned embedded software engineer 20 years my senior, who normally worked for rates I could never afford but enjoyed working with me enough to work for free as a business partner. I was too controlling, too overbearing. I argued with him. It stopped being fun for him, and he politely informed me he would no longer keep working with me. I rushed him because I wanted to build a product demo, but he wanted to build something reliable. I wish I had let him work. The good news is we're still friends.

Around the time I was doing all this, the Facebook movie came out. As I mentioned, Mark Zuckerberg is one of Silicon Valley's success stories, and the dramatic retelling of his story made the startup life look irresistible. This college dropout spent his evenings drinking beer and working at his computer (just like me!) and soon a Silicon Valley venture capital firm was writing him a check for $500,000.

I had no idea how far I was from where he had been, but all the stories of success from Techcrunch, the accelerator Y Combinator, and the Facebook movie combined and fed in to my fantasies that somehow through sheer will and my own brain power I could achieve similar success.

When I was a kid my mother always underscored how intelligent I am. And certainly I did well in school. In 8th grade I got first place in the county math contest and won a graphing calculator as a prize. But I never did meet people as smart as some of the founders I've come across in more recent years. I am still confident that I am smart, but I think a dose of humility could have done me well when I was younger.

Still, I was convinced I could find my success on the funding circuit in Silicon Valley. But after trying to raise funds for my robotics business for six months, I was coming up with nothing. I was running out of money and my product was plagued with serious problems. The motor didn't move well at low speeds, it would overheat, and the latest electronics design had some unknown flaw that caused boards to fail after just 5 minutes of operation. I had a complicated product that really needed an electrical engineer, a mechanical engineer, and a software engineer just to get it working - nevermind the fact that we had no business person to find customers and help define the product specification.

I was running out of money and had no customers. Something serious had to change. And so, never one to give up, I did what Silicon Valley calls a "pivot". I had recently added a wireless controller to the robot I was working on, and was amazed at how cheap the wireless chips were. In my own experience, adding wireless to a product was expensive and frought with difficulty. I wondered - if I threw out all the complicated mechanics of my product and just focused on an easy to use wireless board for hobbyist projects, would that be a simple enough product for me to design myself?

It was worth a try.

I spent the next six months designing prototypes - software, hardware, and tech demos. My girlfriend joined in with me and started writing marketing copy, designing an awesome website, and emailing media contacts who would publicize us when we were ready - we had decided to launch our product on Kickstarter.

It was a very busy time. I barely slept. I lost 20 pounds. I got in a fight with my father after he declined my request to borrow money. I found three friends who would loan me money - to the tune of $10,000 - to keep me going when I had totally run out of cash.

A week before the launch, I still didn't have a good plan for pricing, so I made it up. I knew I wanted the main board to be $20 - even though the final boards weren't designed so I couldn't get quotes. I guessed on prices for all of the accessories too. The project needed a funding goal - the amount that backers have to pledge before you'll get any money. That's a tricky one - if the amount is too low, you can get too little funding to actually do the work you're planning. Set the funding goal too high and it will be too difficult to get the money you need - and many people who see a high limit won't even pledge because they don't think it's worth it! I made up the number - $80,000.

We knew we wanted to offer accessories for the product, so I came up with a few ideas and put them down. We came up with 5-10 different kits people could buy with all this stuff. We had pricing, marketing materials, prototypes, a cool video we had produced with a great video producer, and media contacts at the ready.

Finally, launch came. Thanks to the work of my girlfriend and business partner at the time, we had amazing media coverage on the first day. That Silicon Valley blog I mentioned before - techcrunch? - they had an article on our campaign on the first day. Along with Hackaday, QZ, and Hacker News. We did a reddit AMA. She even got me an interview for a California radio station! We blew past the $80,000 funding goal in just 8 days, and by the end of the campaign in September 2013, we had raised $150,000 from 1600 backers. This little wireless board sold 4000 units in just 30 days.

This is a day many of us who tinker find ourselves dreaming of. We love to make things and we dream of one day making something that give us overflowing bank accounts and acclaim from our peers. And with the help of some dedicated friends, I found myself in just that position. And I felt sick to my stomach.

We'd been working like crazy for months, and with the closing of our campaign our fate was sealed - we'd made a lot of promises and now we had to deliver. I had promised delivery in six months. But what did I have? I had prototypes. The plan was to redesign the hardware before delivery to use an upgraded processor, a better radio, and a slick design.

It would take 18 months before the first final hardware was ordered, by which time I had almost completely run out of money. I had just enough cash to order 1000 units with no accessories and get them shipped out to the backers. I could prove to the world that the product existed, but 3/4 of my customers would have to wait.

By this time I was drinking pretty heavily every night and smoking a lot of weed. As the clock ticked and my cash got closer to zero, I had no idea what I was going to do. I'd only every had one job in silicon valley and I had quit it to start a company that didn't work out, only to pivot, run out of money, and then start another company that was going nowhere. I worked as hard as I could but the stress was killing me. My girlfriend, who had started the company with me, had long since stopped being my business partner. I couldn't afford to pay for her living and we could never figure out how to work together after the campaign ended. I had been so stressed I stopped being a good romantic partner to her, and eventually she told me she was leaving me. She'd been loaning me rent for six months, I never wanted to leave the house, and I was a wreck every day. It hurt more than I can describe, but I don't blame her.

I found myself with no money, no girlfriend, and no place to go. That was and still is the lowest place I ever got in life.

But at the same time, I had been looking for work. On a Thursday evening I went to a local robotics meet up and met someone who ran his own robotics company. They were hiring, and I managed to get an interview the following day. I told him I could only work three days a week - I had a business to run and I needed to figure out how to fund it. Amazingly, he offered to give me a loan to help with that if I would work full time. And so as we walked around the neighborhood near the garage his company started in, he agreed to loan me $55k in addition to a decent salary if I would start work immediately. I thought about it over the weekend, and then reported to work Monday morning. Within a week he'd written me a check.

Retelling this story now has me realizing that perhaps there is more magic in Silicon Valley than I have given it credit for. I was in a very low place, and that offer helped keep my dream alive when I thought it was over.

It was now two nearly years after my Kickstarter campaign had launched. I had a big lump of cash again and a job to pay my bills. I discovered that the $55k loan wasn't enough, and convinced my mother to co-sign on another $18k loan which she had borrowed against her home. I was finally able to order all the hardware I needed to close out the campaign - and between the back rent I owed my ex and the two loans, I was $85k in debt.

The change in my life was significant. I had to move out of the apartment I had been living in for two years - it was more my ex's place than mine. I was no longer working from home either - I had to show up to work every day. I was working as a software engineer, and while I'd been writing software for 15 years by then, I'd never written a lot of it. I quit drinking alcohol and I quit smoking weed. I started riding my road bike and losing weight. I focused on my new job and on my personal health. I didn't do a lot of work for my business - we called it Flutter Wireless - for six months or so. Around October 2015, I transferred $61,000 to a bank account in Shenzhen China. I'd gotten quotes from my manufacturer and was finally ready to order the big batch. I waited months for product to arrive. The accessories came first, since they were easiest to make. But finally by March, huge boxes began to show up at my door. Thousands of units of hardware were finally in hand. It was an incredible feeling!

Suddenly I had a new problem - I'd have to ship all of this stuff! When I ordered the first 1000 units, the manufacturer handled all the shipping. But when the new order came around the manufacturer said they no longer offer shipping services. I'd been working on getting ready for shipment for two and a half years, but it took so long I almost never imagined it would happen. I certainly wasn't prepared. But I did some research, found some nice shipping boxes (which were more expensive than I'd like, I should add) and got ready to ship out the first new orders.

But then I discovered a problem. A big problem.

Flutter Wireless sells wireless boards - little programmable Arduino like boards with wireless radios on them. You can program the boards to talk to each other, so if you wanted to add sensors to your garden or make a remote control robot - Flutter would have you covered.

Only I noticed that a bunch of the boards wouldn't talk to each other. Normally in any Flutter network you have one Master coordinator device that sets everything up, and then a bunch of slave devices that all synchronize with the master. But I found that I had a bunch of boards that would never synchronize to the master! It should be a pretty simple process - program one board as a master. Program another as a slave. Power on the master and the slave, and within two seconds the light on the slave flashes green to let you know it's connected.

Well I had a bunch of boards that would never connect. The light flashed blue and red to tell me it wasn't connected. I started digging in to my software to figure out why this had happened. Eventually I discovered that different boards had different frequency offsets. Each board has a digital radio chip on it that sends and receives wireless signals. But wireless communications uses different frequencies to do different things and if two devices want to talk to each other they need to be using the same frequencies. On each board is a little device called a crystal that creates electrical vibrations at a very exact frequency. The software tells the radio chip what frequency to use and the radio chip uses the crystal as a reference to make sure the frequency is right.

But there's no such thing as a perfect crystal. The way they're manufactured determines how accurate they are. Cheaper crystals are less accurate than more expensive ones, as you'd expect. That's why the bill of materials for Flutter specifies a good quality crystal from a good manufacturer. We list one exact part number from one manufacturer as the part you're supposed to use. And after thorough inspection of my boards, I found that the manufacturer did not use the part I specified, but something else. I emailed pictures of the crystals to the real chip manufacturer and they confirmed - many of the crystals were fake clones and others were simply parts from a different manufacturer. I emailed the other manufacturer and they couldn't confirm the accuracy of the crystals I had.

Flutter communicates using 20 kilohertz wide channels, and some of the crystals had 40 kilohertz of error. When two devices with bad crystals tried to talk, they'd be ever so slightly off - ultimately leaving them on the wrong frequency and totally unable to talk to each other.

It would take months to come up with a solution. We found that the errors in frequency did not change with time or temperature - the differences between units were up to manufacturing differences but whatever error they had would stay that way. There's also an onboard memory on each chip. I worked with a local wireless expert - Earl McCune - who had graciously been helping me on the tough parts of this project for quite some time. He had a super accurate wireless transmitter that we could use to calibrate my boards. So I took five radios and measured the error for each one. Then I programmed them to shift their frequencies just a little - so if one unit was 15 kilohertz too high, it would always subtract 15 kilohertz from the commanded signal.

Then I took my five adjusted boards home, and wrote software that would allow someone to take a box of hundreds of boards and quickly calibrate each unit with the same kind of offset. My chinese manufacturer had a local employee who lived in a nearby town, so I took my 3000 boards to her house, set her up with a computer and the calibration program, and she calibrated all 3000 units for me over a period of weeks.

FINALLY I was had the hardware I'd been working towards. It was now three years since the original campaign. The hardware worked, I just had to ship it.

I would eventually learn that shipping hardware is a pain too. I had about 1000 packages to ship, but the customers had all given me their addresses years before. The software I used to track customers was designed to take their address once, but I now needed them to update their addresses. I had no way of knowing who had done so and who hadn't, so I finally ended up manually emailing them to confirm addresses. I had given the customers a lot of kit options, and then to collect more money I had also given them the option to add in extras. This meant that while about 200 of the orders were the same, most orders had unique options and so had to be stuffed individually.

I hadn't planned appropriately for shipping, and doing it myself was slow. I found that I could move maybe 20 packages a weekend, but it was slow constant work and I hated it. It would take a year of that work every weekend to ship everything.

Eventually I found a local shipping company that would mail out my product. They sent the first shipment around July of 2017, and they've now shipped hundreds of packages. I still have a few hundred more, but I can get 100 shipments set up in two weekends and then have a few weeks free to work on other parts of the businesses before they need more shipments again.

Recently I've started working on other improvements to the product. I added an important feature to the wireless protocol, and I translated the design files from an old design program to a newer open source program I now like to work in. I started working on this product over five years ago, and I'm still not done shipping product to the original backers. It has cost me almost $100,000 of my own money to do this, and I'm still paying off the debt. I had fantasies of investors swooping in with big paychecks so I could work out of a house in Palo Alto with a pool and a barbecue. Instead I found myself drinking to handle the stress of my funds running out.

But fantasies aside, I had another more concrete goal. I didn't have any idea how to run a business, and I badly wanted to learn. I've wanted to start my own robotics company since I was 11 years old, and I still haven't done it. But I have run a successful Kickstarter campaign and raised $150,000 online. I have designed a really nice wireless board that works with Arduino software and has over 1 kilometer range. I started my own open source business and had over $80,000 worth of hardware manufactured in china. I learned the hard way that you need to be careful with your vendors or they'll substitute a critical part for something out of spec. I learned that you have to test your product as soon as it arrives, and not two months later.

I've also learned how much work it all is. I no longer want to sell 10 different kits with all kinds of add ons. I will never again make up my pricing and I will never price things that low. Remember the $20 price point I mentioned? I thought the boards would cost around $12, and I had hoped our $30 premium unit would help compensate for the low profits of the $20 units. But when all was said and done, those boards cost $19 to make. I did all that work to make no money. I'd later learn from some hardware experts that you really want to price goods like mine around 5x the cost to manufacture.

Amazingly, I'd still like to do a Kickstarter again in the future. But I won't do it with a product that isn't designed yet. I'm going to design new products and fund them myself in small batches, selling them on Amazon. If the product works and people like it, then I can consider Kickstarter. And when I do it, there won't be multiple kit options. There will be a single shipping item. I'll already have a thorough test plan in place with documentation for my vendor to follow. If I do a Kickstarter again I'll be able to take the money, send it to the vendor, get tested hardware to my door, and then send it to Amazon for fulfillment.

I lost $100,000 over five years on my first halfway successful venture. But as far as I'm concerned, it's been a success. I had no idea what was going to happen when I started, but I knew I would learn. And learn I have. The business is still going and I still have people excited to use the product. I'm slowly building a brand I believe I can use for future products. I'm considering now a new product I might be able to start selling on Amazon in maybe a year's time under the Flutter brand. It's been in the works for a long time but after all the Kickstarter hardware is shipped I'm going to rebrand Flutter Wireless to Flutter Electronics. My dream is that we'll sell robotics kits. Maybe in another five years I'll get there. When I do, I'll finally have realized the dream I've had since I was 11 - to run my own robotics company.