A Japanese American named Fred Korematsu refused to go to an internment camp. For this he was arrested and convicted.

He sued the government and his case made it all the way to the Supreme Court. In Korematsu v. The United States, the court upheld his conviction 6-3, ruling it constitutional. Those were incredibly dark days for America.

What gives me hope is that American values prevailed. It took far too long but in 1983 Fred's conviction was overturned and in 1998 he was awarded the Presidential Medal of Freedom, the nation's highest civilian honor.

There are now over 1.2 million Japanese Americans and they're a bedrock of society. Many of those who were interned went on to become political, business, and cultural leaders in America.

Norman Mineta was 10 years old when he and his family were forced from their homes and into a camp in Wyoming. He went on to become a U.S. Congressman and the first Japanese American to serve in the Cabinet, as Secretary of Commerce under Clinton and Secretary of Transportation under W. Bush.

When was 4 years old he and his family were interned. He spent the next 3 years in 3 camps, behind barbed wires and armed soldiers. He went on to star in Star Trek where he inspired a generation of kids to look to the stars. He also wrote a children's book "My Lost Freedom" to help ensure we never forget what can happen with a cruel, unjust, and inhuman government. As Takei tells us, it's important to remember the damage that unchecked presidential power has caused this country. 

It's even more important for all of us to, like Fred Korematsu, fight relentlessly for a better future. We must choose hope over despair. If we believe we'll overcome the challenges in front of us, we will. We've been through tough times before. With resilience, we prevailed. And we will again.

In blitz chess each player has only a few minutes to make all of their moves. “Watch your clock” is a phrase you’ll often hear blitz veterans nudging beginners with. Classical chess players – those that are used to 6-hour games – often have difficulty adjusting to blitz. They focus too much on the position on the board and don't pay enough attention to the time left on the clock. In blitz, if you run out of time you lose, even if you're one move away from checkmating your opponent! Classical chess players will sometimes lose badly on time and say something like "but my position was winning!" It shows a lack of understanding of the game. It doesn't matter how winning your position is if you’re badly losing on time. Accomplishing things and the quickness with which you accomplish them are equally important in blitz. 

Startups are like blitz chess. What you achieve is important, but just as important is how quickly you achieve. When a startup raises a seed round they often think about their Series A milestones – what they need to accomplish to unlock their next round of funding. They might set the following goals to unlock their A: “Launch the product, derisk a key piece of the technology, and get at least 3 customers signed for $100k+ contracts.” That's classical chess thinking. Why? No speed goals. In classical chess, players can afford to sink into long bouts of strategic contemplation. But in blitz chess, every second counts. Spending too much time finding the perfect moves can cost you the game because time – not just the position on the board – is an ultimate arbiter. Founders should think just as much about how quickly they can achieve milestones as they do about the milestones themselves. Better goals would be “Launch the product within 4 months, derisk a key piece of the technology within 8 months, and get at least 3 customers signed for $100k+ contracts within a year.”

Why is this important? Because speed unlocks speed.

Investors care about how much progress has been made per time per dollar spent. If you achieve the same milestones in 6 months that it would take an average team 1.5 years to accomplish, investors will be tripping over each other to give you money, helping you accelerate further. If you hit those same milestones in 4 years, you might not be able to raise at all.

Talent looks to see how quickly a team has been moving. Jeff Bezos says that the number one thing that drives away top talent is slowness. When you move fast, it's easier to recruit the best people, which helps you move even faster, creating a virtuous cycle. If you’re slow to reach your goals, good people won’t want to be on your team, slowing you down further, creating a vicious spiral. 

The desire not only for progress but quick progress is rational because speed compounds. The faster you move, the faster you can move. The best indication of how quickly a team will move over the next year is how quickly they’ve moved over the last year.

In both blitz chess and startups, your position means little if you lose track of time. So next time you plan your milestones, don't just focus on what you need to achieve – focus also on how quickly you need to achieve. The time you have on the clock is just as important as your position on the board. Watch your clock.

One of the biggest misconceptions about companies building in atoms and not just bits is that they necessarily have to scale revenue more slowly.

While it’s true that some deep tech companies required years of engineering before being able to sell a complete product (e.g. SpaceX, Tesla), many of the best companies building in atoms scaled revenue extremely fast by launching a product soon after founding. Apple Computer brought in $774k in year 1, $7.9m in year 2, $48m in year 3, and $118m in year 4 after founding. It looks even better inflation adjusted.1

Apple is not alone. One year after company founding in 1939, HP had made $121,092 (inflation adjusted) selling resistance-capacitance audio oscillators. Four years later revenue hit $10m (inflation adjusted).2

AMD was founded in May 1969 and by November had manufactured their first product: the Am9300, a 4-bit MSI shift register. Sales began the next year and 2.5 years after founding AMD’s annual revenue hit an inflation-adjusted $35.6m.3

NVIDIA took two years after its founding in April 1993 to begin selling its first product. Once it launched the NV1, one of the first 3D graphics accelerators, it took only 6 months for NVIDIA to make $2.4m in inflation-adjusted revenue.4 That’s over twice a fast as a top decile SaaS startup takes to get to a similar ARR after launch.5 The next year, NVIDIA’s inflation-adjusted revenue hit $7.8m, followed by $57m in 1997.

These early growth stories stack up well when compared to some software darlings. It took Figma over 3 years to have a live product, ~5 years to start generating revenue, and ~6 years to hit $1m ARR. Loom took ~4 years after founding to hit $1m ARR, Linear took ~3.5 years, and Segment took ~3 years.6 Ramp, one of the fastest growing software companies in modern times, took ~10 months from founding to hit $1m ARR – slower than Apple and at least one Fifty Years portfolio company that’s building in atoms.

Don’t let anyone tell you building in atoms necessarily means slower growth. Some of the fastest to grow startups in history did so by manufacturing advanced physical goods. And when they do scale, they typically have stronger moats than your typical SaaS startup.

Thanks to Gaurab Chakrabarti, Sean Hunt, D. Scott Phoenix, Ela Madej, and Leo Polovets for offering feedback on drafts of this post.

The pace of development in AI has given many the feeling that the ground is shifting under their feet. While incredibly exciting, this has led to a fair amount of anxiety among entrepreneurs who are wondering if there’s any true defensibility in what they’re building. A battle tested strategy in startups is to build a product that’s at least 10x better, 10x cheaper, or 10x easier than what exists while you march toward a long-term moat. But given how quickly AI development is advancing, a 10x product of last month may be obsolete this month. The fear is real.

How fast are things moving? The following big releases all happened In a single week, from March 12 to March 19th:

• OpenAI released GPT-4

• Anthropic released Claude

• Stanford students released Alpaca

• Google released PaLM API

• Google announced Med-PaLM 2

• Google added AI to Workspaces

• PyTorch 2.0 released

• Microsoft Office 365 Copilot

• Midjourney V5 released with nice examples

In a single week, we saw more big announcements in AI than we’ve seen in any other year. Workflows that were previously the sole focus of some startups, over the course of a week, became well done features of companies with massive distribution. Other companies woke up to discover that their product functionality could now be replicated via simple written english.

That week even raised the question of whether foundation models are defensible. A team of Stanford students took a very clever approach of starting with LLaMA 7B, an open source LLM, and fine-tuning it on 52,000 instruction-following demonstrations from OpenAI’s text-davinvi-003. The resulting LLM, dubbed Alpaca 7B, can run on a MacBook and has roughly comparable performance to OpenAI’s GPT-3.5, which relies on massive cloud compute. It took OpenAI 4.5 years and over ~$1B raised to launch GPT-3. Alpaca? Training costs totaled less than $600 (not a typo). It’s a very big deal.

The next week was no less exciting. OpenAI released ChatGPT plugins, potentially obsoleting another round of LLM startups. Google released Bard, its chat interface LLM model. And NVIDIA released foundation models-as-a-service, potentially abstracting away something that previously required many AI scientists.

What a whirlwind. Ideas that seemed solidly defensible even one month ago, might not seem so today. Many companies that were recently the “ChatGPT for X” realized they may no longer have any product advantage after OpenAI Plugins launched. While incredibly exciting, it’s also leading many, including experienced founders and product builders, to have AI anxiety. Some have even attempted to coin phrases for the feeling. Successful founders have gone as far as to suggest, if they were thinking of starting something new in this environment, that they would instead pause and let things settle first.

So where is defensibility in the age of AI? There’s one clear answer. Defensibility in the age of AI can be found by building with atoms, not just bits.

Because robotics hasn’t kept pace with AI, companies that build physically aren’t at risk of being disrupted by AI in the near future. Companies like SpaceX, Cover, or Solugen will gain efficiencies by implementing AI but have little to fear from the latest LLM. But these are not, at their core, AI companies.

What about companies that are AI-first? As building foundation models becomes easier, and as building on top of existing models is sped up, companies that rely on publicly accessible or easily obtained data will likely face fierce competition. Products that are 10x better today may be clearly inferior in a few months. This is unique in the history of technology. Never before have capabilities jumped so far, so fast. Luckily, for AI companies, defensibility can be found in data. 

Every model requires data to train with and then more data to fine-tune. DeepMind’s AlphaFold beautifully illustrates the importance of this data. By elucidating the likely structures of nearly all known proteins and making these predictions freely available, AlphaFold helped biology take a giant step forward. From accelerating basic science to enabling better bioengineering to advancing therapeutics development to helping unlock biomanufacturing, AlphaFold is a big deal.

So why were we able to accomplish so much more with AI in protein structure prediction than other areas of biology? To understand this, we turn to the world of atoms, not bits.

The key enabler for using ML to predict protein structure was in the abundance of high quality ground truth data to train the model on. AlphaFold leaned heavily on two datasets in particular: the Protein Data Bank (PDB) and UniProt.

PDB includes the precise 3-D structure of nearly 175,000 proteins. UniProt is a protein sequence and function database that contains over 200M protein sequences, with over 550,000 of them manually annotated. Building both databases required technology from the world of atoms. The scale, diversity, and precision of the UniProt dataset would not have been possible without next generation sequencing technologies, while resolving the protein structures in the PDB relied on X-ray crystallography, NMR spectroscopy, and cryo-electron microscopy. Without these existing datasets, which required advancements in the world of atoms to collect, AlphaFold simply wouldn’t be possible. 

Imagine these public databases didn’t exist and a startup had developed the only technology that could find the 3-D structure of proteins en masse. No one but that startup would have been able to develop AlphaFold’s capabilities. Quite the moat!

Biology – alongside chemistry – is particularly ripe for advancement with AI because it’s inherently “structured”. Biological reactions happen in largely predictable ways, following the laws of physics, and yet we can’t gain much from the mathematical non-AI based modeling approaches we use in physics. Biology is also inherently highly dimensional in a way that requires increasingly powerful computation to allow us to interpret and understand our data. Unfortunately, there are many areas of biology where the equivalent of  PDB + UniProt don’t yet exist. In these areas, the capability for massive advancements using AI are limited.

DeepMind co-founder Demis Hassabis says “there’s still obviously a lot of biology, and a lot of chemistry, that has to be done.” What’s the key to unlocking similar AI advancements across those other areas of biology and chemistry? Mainly the ability to access large, high quality datasets. If a startup can develop technology that allows for the creation of that sort of data, then it can rocket the field forward while being able to rely on a pretty strong moat. Any startup that can alone access the best data to train and fine-tune with will have strong defensibility. Those with technologies for gathering higher quality, higher resolution, higher dimensional, and/or better annotated data are well positioned to capitalize on upcoming advances in AI. Startups that are able to uniquely generate those datasets in high throughput and at low costs hold the keys to the castle.

One might argue that with enough clever AI engineering or enough compute, the need for high-quality training and fine-tuning data could be circumvented. A counter-argument here is OpenAI’s now abandoned effort in robotics. OpenAI co-founder and CTO Ilya Sutskever was recently asked about why they couldn’t make enough progress. The issue? Not enough quality data. Ilya said “back then it really wasn't possible to continue working in robotics because there was so little data. Back then if you wanted to work on robotics, you needed to become a robotics company. You needed to have a really giant group of people working on building robots and maintaining them. And even then, if you’re gonna have 100 robots, it's a giant operation already, but you're not going to get that much data. So in a world where most of the progress comes from the combination of compute and data, there was no path to data on robotics.” OpenAI co-founder Wojciech Zaremba, who was the robotics project lead, said at the time “it turns out that we can make gigantic progress whenever we have access to data” and explained that it therefore made more sense to focus on the “plenty of domains that are very, very rich with data.”

If a startup can quickly, cheaply, and uniquely generate the data needed in a domain to train and fine-tune AI models, they can unlock incredible value. The best way of ensuring that you have proprietary access to a dataset needed to train a model is to generate it yourself. Building and working in atoms can help. 

If your technology is built in atoms and lets you collect high quality data in high throughput and at low cost, then rapid advances in AI should excite you, not scare you. In that case, every leap forward in AI, rather than potentially obsoleting your product or company, simply helps you derive even more value from your unique dataset. This isn’t just true for biology. This sort of technology can come in many forms: assays to see new biology, space-based imaging technologies, new sensors based on advancements in materials science, or even an ability to synthesize a critical component in a design–build–test–learn cycle.

Companies with technology that allows them to uniquely generate the data needed to train and fine tune models are well positioned to create enduring value in the age of AI. The best AI companies may be those building in atoms and not just bits. Create the data, control your future.

Thanks to Ela Madej, Josie Kishi, Elliot Hershberg, and Gaurab Chakrabarti for offering feedback on drafts of this post.

The exact same trend that happened in software over the last decade and happened in hardware over the last 5 years or so is happening in biotech now. A smart software developer can build and launch a web or mobile app and get paying customers for under $2,000. The same will soon be true for biotechnology.

Here are a number of trends that are fueling the steep drop in the costs of founding a biotech startup.

1) Biohacking spaces & shared wet labs.

Research intensive companies used to have to either grow out of a university lab or spend quite a lot of capital on their own lab. Now, they can get off the ground in biohacking spaces like Counter Culture Labs or Biocurious and, once they’ve matured from project to company phase, cut costs by joining a shared wet lab like Harlem Biospace or Qb3. There, in exchange for a monthly fee resembling that of a co-working space, they have shared access to cutting edge equipment needed to incubate cultures, grow tissues, and much more.

The Real Vegan Cheese project,3 born out of Counter Culture Labs, is an example of what can be done for very little money. Through genetic modification, they’re very close to getting yeast to produce milk through the process of fermentation instead of alcohol. They funded the project with ~$37,000 that they raised on Indiegogo.

2) Better tools.

The primary thing that dropped the cost of building software companies was access to better and cheaper tools and services. The quintessential example is Amazon Web Services, which allows developers to host their servers in the cloud, meaning they no longer have to invest in expensive and difficult to configure hardware (servers).

The same thing is happening in biotech. Tools like Transcriptic are essentially AWS for the life sciences – a cloud laboratory accessible through an API. OpenTrons offers affordable robots for wetlab automation. Mousera is Heroku for mouse studies, and helps automate much of that process. Google launched Google Genomics, which allows researchers to run complex computational analysis and store their data in the cloud. Benchling drops the cost and speeds up the pace of research by automating many tasks (like counting bacteria colonies) that were previously manual.

Other tools are helping solve the problems of unused equipment capacity. When companies had to buy their own servers, they would often sit at 15% capacity for most of the time – a terribly inefficient setup. As things moved to the cloud, the entire system became more efficient, dropping costs for everyone. The same thing is happening in biotech. Companies like Science Exchange allow researchers to outsource their experiments to university and commercial labs that have excess capacity. On the site whole genome sequencing goes for around $1,000. Others like Synaptic allow universities that have expensive research equipment sitting idle to more effectively lease time on the equipment to researchers that need it – be they academics or entrepreneurs.

Other tools are being developed that will eliminate the need for animal (and maybe even human) testing of new pharmaceutical treatments. Emulate, for instance, is a DARPA supported company that is creating living systems on a chip that can be used to test drugs with more predictive results than animal testing. In-vitro techniques are being used to grow human skin which can be used to test new topical treatments.

Tools like these allow researchers to cut both the time and capital cost of research oftentimes by orders of magnitude, and the tools space is only just heating up.

3) “Open source" biotech.4

One of the factors that contributed most to the cost of software falling was the advent of open source software. Software companies make use of freely available tools – from server technology to front end frameworks – that speed up development time and reduce overall costs. The same thing is starting to happen in biotechnology.

There are more and more freely available open source datasets that allow biotechnology researches to skip the collection step, which can often be incredibly costly and time consuming.

The most famous and the largest such dataset is the Human Genome Project, which mapped and made publicly available a complete genetic blueprint for building a human being. Another model example is the International HapMap Project, an open source catalog of genetic variants that occur in humans, on top of which researchers can find connections between specific genetic variations and certain diseases. Amazon freely hosts human DNA sequence datasets, microbiome DNA sequence datasets, and many more collections of data that people can experiment with easily and cheaply.

The increasing abundance of freely available life science data will enable biotech entrepreneurs to move faster, cheaper.

Simultaneously, the open access movement in biomedical publishing is gaining steam. Organizations like PLOS and PeerJ are leading the way here. A world where life science scientific papers are widely available for no or little cost is within sight, which may help spawn a new generation of scientist-entrepreneurs.

4) Software + biotech.

More and more research that previously necessitated capital-intensive labs can now be done computationally. The best analogy here is the impact CAD tools had on the process of developing a hardware product. CAD allows hardware designers to test a multitude of different design variations without incurring large time and capital costs. Software is transforming the life sciences in much the same way.

Software and AI are inserting themselves into more and more parts of the processes of biotech, reducing costs and timelines dramatically. One of my favorite examples of this is the startup Atomwise, which uses deep learning to discover promising drugs, making one of the most costly parts of the drug discovery process more capital efficient.

These techniques are even more important in drug discovery than in hardware. The chemical space in drug discovery – the ensemble of all possible molecules that are potentially interesting – has been estimated at 1060. This is a tremendously large number that is only truly navigable through computational means. By enabling pharma companies to more intelligently choose which candidate drugs to bring to trial, tools like Atomwise will speed up the process and reduce costs in a significant way.

5) New sources of funding.

Traditionally the only checks available for biotech commercialization were very large. There haven’t been “seed stage” biotech VCs as we understand the term. This means that it’s hard for “hackers with a crazy idea” to test some of their assumptions unless they could self-fund. This too is starting to change.

The exact same thing happened in hardware. In 2008 when fitbit raised only $2M in their first round of funding, many were very skeptical that a company making a hardware product could do very much with such a small amount. Last month, they IPO’d after having raised only $66M in total.5

We’re starting to see more seed-stage VCs focusing on biotech, and it’s now possible for early stage companies to enter biotech-specific incubators like IndieBio that offer both small amounts of funding and access to expensive wet lab equipment. Y Combinator, the world’s most prestigious startup incubator, is funding more and more biotech startups. In the last graduating batch, there were 18 biomedical companies, and these numbers are likely to only increase. This trend is both a result of some smart investors recognizing the above trends and will also help fuel them, by allowing biotechnologists to raise a small amount of capital to reach key milestones and prove some assumptions in a low cost way.

Sites like Experiment offer nontraditional sources of funding for researchers in the early stages. One example of what the future might hold is a project aimed at discovering a new treatment for Ebola, which recently raised $140,000 in crowdfunding. Before launching the $140,000 campaign, they first launched a $5,000 fundraising campaign, which gave them enough capital to prove some assumptions cheaply by running experiments through Science Exchange. Only after getting positive results did they decide to raise more for the next phase. This lean setup – raising a small amount to test assumptions cheaply then repeating the process with increasing amounts of capital – will likely become more common over the next decade. This mirrors what happens now with software but has only recently become possible in biotech because of the trends mentioned above.

Conclusion

From consumer to SaaS to hardware – it’s always interesting to imagine what form the most exciting innovation will take in the next decade. For the reasons above, I can’t stop thinking it’ll be biotech.6

We’re not quite there yet, but in the near future biotech entrepreneurs will able to get their companies off the ground from their garage, funded off their credit card. It’s interesting to imagine the wave of innovation this will make possible.

Notes:

1 The costs of sequencing the human genome are falling far faster than Moore’s Law would predict, as shown in this chart.

2 The one notable exception here is pharma and specifically the cost of taking a new drug to FDA approval, which has jumped from around $800M in 1999 to $3B to $4B in 2012. That said, some of the trends mentioned in this post should play a big role in dramatically cutting those costs.

3 Wired published a great article on the Real Vegan Cheese project that’s worth a read in full.

4 ”Open source” is not the most technically accurate term for the phenomenon I’m describing, especially as it’s used in the software world. It is close, though, and there doesn’t seem to be anything better. Yochai Benkler uses the phrase “nonproprietary peer-production of information-embedding goods”, which is technically more precise but doesn’t make for a good blog post header.

5 The Fitbit IPO returned one of their early investors (Foundry Group) over $1B. When an industry is getting increasingly capital efficient increasingly fast, it certainly pays to spot the trend and put money to work early.

6 Biotech is a broad field. While I think costs will drop across the entire market and many sub-areas will see much innovation in the next decade, synthetic biology in particular seems to offer the greatest promise.

Thanks to Max Hodak, Matt Owens, Cindy Wu, and Alexander Levy for sharing their insights on these trends and to Ela Madej for reading a draft of this post.

Last year, my startup Amicus nearly died. Running out of cash, we needed to find a way to cut every cost possible. We had to let go of some amazing people, cancelled every non-essential service and perk, and asked companies for free months and refunds for time when our usage hadn’t been as high as expected.

Some companies said no. Others bent over backwards to help us survive. Some gave deep discounts, others free months, and some even refunded us for past months when our usage didn’t match the plan we were on.

I have trouble imagining companies like Comcast or Bank of America or ADP giving refunds and discounts when they’re not contractually obligated to, just to help a struggling startup push through rough times. And yet a dozen startups for whom every dollar is often important jumped at the opportunity to help.

Without the support of these companies, we might not have been able to survive. So to all of them, thank you. Not only do they offer products that we appreciate at Amicus, but I can attest that they’ll be there for you when you need them most. These startups truly care about their customers.

Lever

Segment.io

Travis CI

NiftyQuoter

Zirtual

Zendesk

Basecamp

Github

Twilio

SimplyInsured

Safari Books

And thanks also to these individuals:

Bill Schnoor at Goodwin Procter for being exceptionally accommodating with fees; Pedro Torres-Picón at Quotidian Ventures / qLabs for giving us free office space; and Chris DeMayo at Withum Smith & Brown for digging us out of a messy tax situation while keeping costs low.

I hope those of you reading this seriously consider giving these companies your business.

Thanks to Ela Madej for reading a draft of this post

More and more, I hear the question “what do you hope to achieve by founding X” answered with “we want to make X a billion dollar company”. That’s often followed by some mention of “disruption” or “innovation” or even “impact”, but the core purpose of these founders is clear – to make boatloads of money. Aside from the fact that there’s already an industry for those primarily interested in accumulating wealth, that’s a recipe for startup disaster.

While it appears the vanity $1B mark matters even to some entrepreneurs en route to big success,1 the harms in striving to be included in the “billion dollar club” manifest most in aspiring entrepreneurs – those founders just setting off on their startup adventures.

The best startups are born because the founders see a problem in the world that no one else is fixing, and decide to build a solution themselves. When you ask the best founders what they hope to achieve with their companies, you get answers like “I want online payments to stop sucking” or “I want the world to be a more connected place” or “I want people to have a voice in the political process”.

The best founders first and foremost want to impact the world in a way they believe is meaningful and if they get rich doing so, great, but that’s rarely the goal.

If you’re an aspiring entrepreneur, I urge you to focus on a solving a problem you feel is deeply important. When things get really tough, a meaningful mission will help push you through. When you’re trying to convince the best and brightest to join you, working on impactful things will help.

And if the chances are that you’re going to fail anyway, you might as well fail attempting something meaningful.

Discuss on HN

1 There are 22 startups at the “billion dollar” valuation mark, but only 6 at the 1.1 B mark. I bet there are far more at the 1B mark than at the 900M mark as well.

Thanks to Freddie Andrade and Ela Madej for reading drafts of this post

I had hit rock bottom.

In The Hard Thing About Hard Things, Ben Horowitz says “nearly every company goes through life-threatening moments… it’s so common that there is an acronym for it, WFIO, which stands for ‘We’re Fucked, It’s Over.’” He estimates that most companies go through at least two, and sometimes over a dozen WFIOs in their lifetime. This was our first WFIO, and it hurt even more because things had been going so well. We’d made incredible progress after founding Amicus three years ago to help nonprofits do what they do better – we graduated from Y Combinator, raised nearly $3.8 million in venture capital, grew to a team of 15, and today can call some of the biggest nonprofits in the country customers and fans of our products. Having survived our first WFIO, we’re as dedicated as ever to fulfilling our mission of empowering people to organize around the causes they care about.

Entrepreneurs often write about what’s going right, but too rarely write about what’s gone wrong. One of the big values of Y Combinator was that we were able to hear strictly off-the-record stories of many successful startups’ WFIO moments. It’s a shame more entrepreneurs don’t talk about their most difficult moments publicly, because it paints a distorted picture of what startups are. A lot of the mistakes I made were avoidable, and I hope to shed light on them through this post to help others steer clear of those pitfalls. We have Amicus shirts with “Make Mistakes” emblazoned across the front. They’re meant to encourage people to take risks, be bold, and embrace small failures – after all, if you’re not making mistakes, you’re not reaching your limits. But these are mistakes you shouldn’t make.

Mistake 1: Taxes

I hired our first accountant this year. I knew our books hadn’t been well kept and was pretty sure we owed some state taxes. He determined we owed about $5,000 in state and local taxes, which we quickly paid. But shortly after, he noticed there hadn’t been any payroll tax payments coming out of our bank account.

Impossible, I thought, Bank of America’s payroll system automatically withholds payroll taxes every month, it’s all automated. But it wasn’t. There was a single “submit tax” button in a separate part of the Bank of America website that had to be clicked to actually pay the taxes. And because the IRS had our old company name (“BlueFusion”) and address on file, we never got any notices. And so I learned that we hadn’t been paying payroll taxes for almost 3 years – a particularly painful thing given I believe in taxes as a means of giving back to society.

When all is said and done, cleaning up this mess (the taxes, the fines, the legal fees, etc.) cost the company over half a million dollars. And we had to pay the vast majority of that to the IRS immediately. Needless to say, such a huge financial shock severely hurt our runway and triggered many difficult changes. This hurt even more because I’d been paying myself one of the lowest salaries on the team to maximize our runway and support our mission, and with this mistake did just the opposite.

Lesson learned:

We should have followed operational best practices, even in the early days. We used Bank of America’s payroll system to try to save costs, but our tax issue could have been prevented by using a third party payroll provider that automatically submitted payroll tax payments. It also would have been solved by proper budgeting. Doing that, we would’ve noticed the tax discrepancy in a matter of months, instead of years, and benefited from the many other upsides of budgeting as well. We also should have worked with an accountant to keep our books in order. Financial operations aren’t my forté. I have little tolerance for paperwork, forms, or bookkeeping, and so had been putting off some operational tasks for too long – knowing this, I should have outsourced this as early as possible. We now make use of a bookkeeper, an accountant, and all payroll taxes are paid automatically. Don’t cut corners here.

Mistake 2: Poorly defined co-founder relationships

The first version of Amicus was built by two incredibly talented developers at Yale who chose to finish their degrees rather than pursue the startup full-time. We recruited a CTO and then a close developer friend of mine joined as employee #2. After some time, I decided to give them each “co-founder” titles, though since we already built a product, raised investment, and had paying customers, our equity split wasn’t even.

We applied to and got accepted into Y Combinator, but after we graduated, the relationships started to cause tension. I often felt like I was pouring more of myself into the company than they were and wasn’t able to share founder burdens, and they often felt I didn’t include them on important decisions and didn’t share enough information with them. One minute, they felt like co-founders; the next, they felt like employees. Confusion and resentment grew and communication suffered.

Lesson learned:

If I communicated earlier with my former co-founders to establish set roles and expectations, a lot of the difficulties we encountered would have been avoided. I should have flushed out a mutual understanding of how decisions would be made, what information would be shared among the founders, and what level of commitment everyone was willing to give. I’d even suggest writing these expectations in a formal agreement.

Just as important are frequent, honest checkins to see how everyone is feeling on an interpersonal level. Founder breakups are one of the most common reasons companies fail. By getting on the same page early on and ensuring good communication, we could have avoided the confusion and resentment that became such a volatile force.

Mistake 3: Not being explicit about hacks

Another area where an honest and upfront discussion is warranted is around what systematic hacks are acceptable and what level of risk everyone is comfortable with. Paul Graham looks for founders who are “naughty,” but what level of rule-breaking is everyone comfortable with? I never had a conversation with my co-founders about this and it ended up causing problems.

In the early days I would often let potential customers think we already had a feature they wanted and, if they signed, would come back to the team and say “we’ve got to build this before they launch!” No harm, no foul, I thought, so long as we knew we were able to build the feature before they started using the product. This is a tactic commonly suggested by lean practitioners. My co-founders, though, would often frown on this behavior, worrying it was unethical, causing a huge amount of tension to grow beneath the surface.

Lesson learned:

I should have had conversations with my co-founders about what rules they were comfortable breaking. Testing product demand with a fake landing page that collects credit card information? Manufacturing a sense of urgency when fundraising? Pushing the limits of the law, which we never did, but which many innovative companies such as Airbnb and Lyft are doing? If we had these conversations early on, we could have avoided culture clashes (and worse) later on.

Mistake 4: Going it alone

Too often while running Amicus it’s been my first tendency to try to solve big problems alone – a tendency many CEOs have. My partner recently offered me some unsolicited psychoanalysis on this point. I grew up without a father in my life, and after an explosion that disabled my mother, our roles were reversed and I became her caretaker at the age of 12. From a young age I’ve been figuring things out on my own, and never learned to lean on the guidance of elders. I’ve developed some bad habits in the process. My go-it-alone attitude contributed to many of the mistakes that have been made.

Lessons learned:

I built an amazing team around me over the last three years, and I should have empowered them more to help solve the big problems and take advantage of the big opportunities. A few smart minds trying to solve a problem is almost always better than one mind. And the more you bring teammates in on problems you’re facing, the more ownership they feel over the organization and the more they trust in your leadership. The same goes for investors and fellow entrepreneurs. Moving forward, I’ll be reaching out to investors, teammates, and my network more frequently when we face problems, and lean on them more when we’re presented with opportunities.

Mistake 5: No outside board members

This is related to the go-it-alone mistake, but it warrants its own mention because I think it’s a mistake I see a lot of first time entrepreneurs making. While it seemed like a fundraising victory at the time, not inviting an investor to sit on our board during our first two rounds of financing was a mistake. While fundraising, I maximized for valuation and control, instead of maximizing for value-add. The worst thing about this is that I raised from some amazing investors who would make excellent board members.

Lesson learned:

The debate about whether a seed stage company should have a founder-only board or a board with an outside investor director is still raging. I know that for me, as a first time founder, the many upsides (mentorship, a cadence to the business, outside perspective, pattern recognition, real investor buy-in) would have outweighed the downsides (loss of control, investment of time in board management).

If I had to do it over again, I would have chosen an investor who was smart, cared about our mission deeply, and with whom I got along very well, and invited them to serve on our board. Had I done this, the tax issue almost certainly would have been avoided, and I’m sure a number of smaller mistakes would have as well. An experienced board member who cares about you and your company can use their pattern recognition to help you steer clear of all sorts of pitfalls. I believe so strongly in this that I’m voluntarily creating a board seat for one of our current investors.

Mistake 6: Poor investor relations

When you’re running a company, there are always 100 things that need to be done, and enough time to do 70 of them. One thing that I let slide was investor updates. There were a couple investors I talked to regularly, but by and large my communiqués were scarce, and normally were sent when I needed something. This meant that when the crisis struck, it took much longer for investors to be able to help than it needed to. I had to fill many of them in on the past year of the company, instead of having to fill them in on the past month. And as you can imagine, having the first substantive update you hear in a while be about an impending crisis is not the most inspiring thing. So while our investors have by and large been quite supportive, it took longer for them to get engaged because they’d been kept out of the loop for so long.

Lesson Learned:

I should have talked with our investors more. I shouldn’t have only pinged them when we needed something, but should have kept them updated on the progress we were making (or not making) on a regular basis. I owe it to my team to keep our investors close and I owe it to our investors to keep them updated – after all, they’re taking a huge financial risk to help us achieve our mission. Aaron Harris of Y Combinator wrote a great guide to investor updates which I find very helpful.

Mistake 7: Telling a half-truth

Lastly, this has been one of the most difficult times of my adult life, and has forced me to go through a painful amount of introspection. Through this process, there was one example of rule bending I wish I could take back. Before Amicus, I was enrolled in the bachelor’s degree program at the Extension School at Harvard University – a small program with a few hundred graduates a year. At the time, I considered it a good hack, as I was getting, as the New York Times put it “Harvard at a fraction of the cost.” I even played for the Harvard Chess Team, including traveling with them twice for competitions in China.

Eventually I dropped out to pursue the startup full time. Partly for convenience and partly to make for a good story, I would often say I dropped out of “Harvard University.” The fact that this was technically true allowed me to rationalize the statement to myself. But in saying it this way, I gave people the impression that I had dropped out of Harvard College (à la Zuckerberg and Gates), which was not the case.

At some point a friend called me out on this, so I started being more clear. It felt good to do so, and the press coverage of Amicus that mentioned I went to the Extension School wasn’t any less compelling for it. But the damage was done, and I often fell back into old habits. To cause people to think I didn’t get a degree from one school at Harvard when, in reality, I didn’t get a degree from a different school at Harvard is the height of absurdity. I’d sacrificed my credibility for little benefit other a sense of false prestige and a sexy line in a press story.

Lesson learned:

First and foremost, I’ve come to realize that “technically true” is a terrible ethical measure for a (non-technical) statement. I should have held myself to a higher standard. I’ve also learned the importance of overwhelming honesty. The phrase is borrowed from the excellent book The Transparency Edge (worth a read in full). In the same way a product builds up technical debt with every piece of hacky code, a leader can build up a sort of managerial debt with every fib, embellishment, and exaggeration. They all serve to undermine credibility. And without credibility, you cannot lead. Moving forward, overwhelming honesty is the name of the game at Amicus.

Other mistakes:

Over the three years I’ve been running this company, I’ve made a number of other mistakes as well (burning out myself and my team, on-and-off micro-management, an occasional lack of empathy, bad hires – the list goes on), but the ones highlighted above are what led most directly to the crisis we now face. Each of the other mistakes probably warrants a blog post as well.

This past month I’ve had to part ways with team members I loved working with. We don’t have as many resources as we used to. My co-founders and I shared a passion and a vision for making the world a better place by empowering nonprofits, but now they’re gone and we’ve taken a major step back from accomplishing our goal. Realizing that this is all my fault has been tough to deal with emotionally.

That’s taught me one last lesson: that in tough times a support network is invaluable. They say you only really learn who your friends and supporters are when shit hits the fan, and that’s certainly been my experience. Some of our investors saw these mistakes and threw their hands up in despair – and who could blame them. Others chose to roll their sleeves up and help clean up the mess.

I’m proud of the Amicus team that remains, those who are fighting through the tough times because they believe in our mission. And I’m humbled that they still believe in me. I’ve been most surprised by the outpouring of support from other entrepreneurs (mainly from the Y Combinator network) who heard we might be going through tough times, and reached out just to let me know they were there for me. I couldn’t be more grateful for the investors, teammates, and friends that have supported me through this.

Moving forward

It’s been a difficult month. Paul Graham’s essay on How Not To Die has been an open tab in my browser for the last couple weeks. And I now truly empathize with Ben Horowitz when he writes about The Struggle. But we carry on. The mission of Amicus is too important. We’re trying to upend the power structures that exist in society by giving people the tools they need to organize around the causes they care about. We’re not going to let a few bumps in the road – even if they’re big ones – get in the way of that.

I stayed up all night recently reading about some of the great entrepreneurial turnaround stories. It’s shocking how many of the companies we all know and respect were months, weeks, or even hours away from death. FedEx, Evernote, Intuit, Zappos, Airbnb – I’m endlessly inspired by those founders who faced near collapse but simply refused to give up. I hope one day the Amicus story can inspire someone the way these stories inspire me.

To summarize:

1. Pay attention to financial operations from the early days. Make a budget.

2. Be explicit with your co-founders at the get-go about decision-making, distribution of information, and level of commitment. Formalize this in a written agreement.

3. Have conversations with co-founders and teammates when they join about what rules you’re comfortable bending and what hacks you’re comfortable implementing.

4. Don’t be a lone wolf. Lean on the experience and smarts of your teammates, investors, and mentors to help solve the tough problems and take advantage of the opportunities.

5. If you’re a first time entrepreneur, invite an outside director to sit on your board when you raise money. The upsides greatly outweigh the downsides.

6. Keep your investors posted on your progress, be responsive to their requests, and lean on their guidance.

7. Be overwhelmingly honest with your stakeholders (team, investors, customers).

8. Build a support network of fellow entrepreneurs when the times are good, because when the times are tough their support is invaluable.

These are painful mistakes I’ll never make again. I hope by hearing this story you’ll avoid making them too.

Thanks:

1. Chris DeMayo did amazing work helping us get our books in order and fixing our tax situation. If you need an accountant, I couldn’t recommend anyone more highly.

2. Ashu Desai, Mattan Griffel, Henry Xie, Dan Friedman, Joel Califa, Paul Cretu, and Ela Madej for reading drafts of this post

3. The YC network for being so supportive through this

I recently got back from a five day holiday vacation to the Puerto Rican island of Culebra, planned last minute because I felt dangerously close to burning out. Every time I talked to anyone about the trip, I included the disclaimer “it’s only my third break since Amicus was founded”. Every time. It was only once I was lying on a beach that I realized what I was doing: I was making excuses for taking a break because I felt guilty. To my teammates, to my friends, to fellow entrepreneurs. I might as well have been saying “vacations are for the weak”.

Professional runners take long breaks between marathons. They make no excuses for this, and no one judges them for it, because everyone knows that rest and recuperation is an essential part of being a pro athlete. The same is true for entrepreneurs (and everyone, really). Preventing burnout is part of your job. Staying well rested is part of your job. Sleep and exercise help, but occasional extended breaks are essential too, and their benefits on creativity, productivity, and happiness are well documented.

It’s time we stopped making excuses for rest and relaxation. Doing so is not only bad for you, but sends the wrong message to the rest of your team. So next time you’re planning a vacation, announce it with pride.

Just last night, an Internet company (Netflix) won three Emmy awards. And while the top prize went to a show from a more traditional network, the producer of that show credited Netflix for its success during his acceptance speech. An industry that has long resisted disruption from the tech world instead lauded an Internet company on stage, and offered it their most prized honor.

Last week, a video game (Grand Theft Auto V) reached $1B in sales in just three days. It achieved that milestone faster than any movie, any album, any book. A piece of programmable content is now the fastest growing entertainment property in history.

And at Sundance this year, Kickstarter funded films won 6 awards. In fact, a full 10% of the movies selected were Kickstarter backed. In an industry where the big studios have long decided what films got made, and which won distribution, the process of funding production is now being democratized through technology.

There is still a long way to go – the music industry, for instance, suffers from a stifling lack of innovation – but 2013 certainly seems like a turning point in entertainment. I can’t wait to see what’s next.

Soon I was having trouble dealing with the stress of Amicus’ ups and downs. I had trouble sleeping and even started snapping at my teammates. It got bad enough that I turned to mentors for advice on what to do. I was surprised by the feedback I kept hearing: “exercise more”. So I started irregularly hitting the gym. I didn’t make a real lifestyle change, however, until Amicus got accepted into Y Combinator last year.

Our first day of YC, PG stood up in front of the batch and said:

If you’re ever unsure if you should be doing what you’re doing during YC, ask yourself this question: ‘Am I building our product? Am I talking to users? Am I exercising?’. if you’re not doing one of these things, you’re doing the wrong thing.

This drove it home. YC is a 3 month sprint when entrepreneurs are supposed to focus on nothing but their startups, and here was PG saying we should all make time to exercise – that tending to your body is essential to being a successful founder.

Yesterday Foursquare alerted me that I had been to a gym at least once a week for the past year. Not making time for exercise now seems akin to cutting out food because you have no time to eat – “gotta tend to my startup!” If founders hustling through YC can find time to exercise, so can you. Join a gym. Bike to work. Go rock climbing. Make these things perks so your teammates exercise too.

And remember, exercise makes you a better entrepreneur.

“We can still learn from history the invaluable lesson that an enormously powerful and profitable evil can be overcome.”

–David Brion Davis

Entrepreneurs have an inherent distaste for entrenchment, and seek to disrupt it wherever it exists. Because of the history of Silicon Valley, entrepreneurs are typically focused on market entrenchment. The organizations being disrupted are market goliaths – the Microsofts, IBMs, and Oracles – that use their connections, capital, and incumbency to stifle innovation. In our world, those that bring about this disruption are heralded as innovators and held up as exemplars.

It’s time we expanded our field of vision. There is another category of powerful entrenched interests in our midsts – another flavor of anachronistic goliath. These kind don’t look to protect their market caps or their cash flows; they look to protect their influence and their votes.

Yesterday we saw a grotesque example of this kind of entrenchment, as a minority in the U.S. Senate blocked legislation that would have required background checks on all gun sales, instead of only registered broker sales as is currently the case. Before it was introduced, every controversial measure was dropped from the legislation. It was so uncontroversial that nearly 90% of the American public supported the reform, including 85% of gun owners.

But in stepped the NRA, the Comcast of the political world, to kill the bill. Despite this legislation having the support of a majority of Senators, and nearly 90% of the public, it didn’t even get an up or down vote.

Whatever your views are on gun control, it’s hard to see yesterday’s events as anything but an example of an entrenched political interest using its power (money, exclusive relationships, incumbency) to maintain the status quo (sound familiar to any startups out there?). Former Congresswoman Gabby Giffords writes about this far more eloquently than I can in yesterday’s NYT.

This sort of political entrenchment is legion: big oil kills renewable energy investments, certain religious institutions kill marriage equality, content producers kill internet freedom – it goes on and on, and crosses party lines.

It’s time our entrepreneurial community widens our field of vision and takes on this problem. It’s time we celebrate not only market disruption, but political disruption. This is why we founded Amicus – to give those like Gabby Giffords the tools they need to disrupt the powerful entrenched interests standing in the way of progress. But Amicus is not enough. We need to see technology enable even more disruption in this space.

This work is rewarding. When you disrupt political entrenchment, great things can happen. Slavery can be abolished, women can gain the right to vote, SOPA can be defeated. For instance, Amicus helped turn the tide on marriage equality by helping advocates overcome the entrenched minority standing in their way. And it feels amazing.

If we create the tools that enable political disruption, innovation will flourish, and nothing will stand in the way of progress. So let’s get hacking.

If you’re interested, join us in achieving this mission.