Everything I have learned about startups in my first year in silicon valley

May 28, 2024 marked one year since I dropped off my boxes at the Atlanta Goodwill on Marietta Ave and took a one way flight to SFO. I didn’t know at the time how much my life would change. I landed in SFO a struggling new grad startup founder with a product that no one wanted to now running, if I may say so, a fairly promising and, critically, growing startup. While we are in the nascent stages of our existence, I think there is at least some value in the lessons I have learned.

This is a brief summary of everything I have learned in my first year in silicon valley.

I learned what a startup is.

“Startup” is a strict subclass of “business”.

A business is an entity which gives people something they want but don’t have or know how to get, in exchange for some type of value, nearly always in the form of money. Good businesses provide a lot of value to a lot of people, and collect a large percentage of the value they create. They must also do this while spending less than they make. In essence, a business needs only one thing - a paying customer. Not a product, not a service, not employees, not even an office.

All this might sound too obvious to say out loud but surprisingly, I wasn’t alone in being late to this realization. There is a huge stress on startups that focused on something that was not having many happy paying customers. Moonshot labs, social networks, really any startup you heard of from a TechCrunch fundraising announcement often give people the false impression that a startup are not simply a type of business, and that all the rules that apply to businesses don't apply to startups.

Startups are businesses that commonly have the following characteristics:

• Technology: They employ technology to allow them to scale - quickly increase the number of customers they serve by driving down the cost of goods, or using new techniques to improve the experience their customers have, so much so, that they are willing to pay more. An example is Uber being larger than any traditional taxi company because software allows them to cheaply recruit drivers, register new riders, and do the job of a dispatcher at scale.
• Growth: Startups have usually identified a need or problem that people have, and/or a novel way to solve it better or more efficiently. They must grow extremely fast to prevent others from capturing the alpha.

When people in silicon valley use the word “startup”, they usually mean “venture-backed” or “venture-scale” startup. These startups are ones that raise some type of outside money from investors for a percentage of their company’s ownership to fund their initiatives. This can include companies that are solving expensive technical problems like supersonic jets, or companies that can use the funds to grow much faster, which their survival might depend on.

To be a venture-backable company, your startup must have two characteristics:

1. If it succeeds, it must become huge. This is because investors make early, high-risk investments which only make sense if there is the possibility of outsized outcomes. It can be quite unintuitive how low the odds of success are predicted to be, and how large the potential size of the outcome is predicted to be.
2. It must be a company that can grow extremely fast given sufficiently large amounts of money. By growth, I mean net operating income.

I learned how to pick what to work on

There are a variety of opinions on this but mine generally boil down to supply and demand. What is a problem with a small supply of potential solvers - that very few people know about or have the skills and willingness to solve, and that there is disproportionately more demand for?

We started working on Greptile because we had faced the problem ourselves and understood it better than most people (naturally, since it is a dev tool and most people don’t code), and we had disproportionately more skill than most people to solve it (we had worked with RAG systems and built on LLMs just about as long as anyone). Since we had faced the problem ourselves, we knew that a lot of people had this problem too, and that somewhere in the vicinity of this startup idea was a billion-dollar company.

I learned how to raise money

This is a subject that is very widely written on. The framework I have arrived it is "don’t seek, attract". That is to say, just build a great company and if you do it out loud, investors will find you.

This is different from customers, whom you have to seek out yourself. The difference is that investors’ job is to find you. Without you, they would have nothing to fund and would cease to exist.

“Don’t seek, attract” works on both the macro and the micro. On the macro, it means instead of seeking out investors, just build a great company - which means one with growing revenue, solid unit economics, and happy customers. They will find you. On the micro it means when you’re in the room with an investor, be self-assured, confident, and talk about how great your company is. This is about you, not them. The hot guy at the bar isn't trying to hit on everyone, he's just, is, hot.

I learned about geographical advantages

Many people would consider this controversial, and maybe I have chugged the YC kool-aid on this, but I think your startup should be in the bay area. The reason is very simple - this is where the best founders are, so if you’re here, all your friends will be amazing startup founders, and they will encourage you to work harder, you will learn a lot from them, and you will be surrounded by people who understand the thing you’re trying and the thing you are going through. This is hard to overrate.

For us, moving to SF showed us where the bar was. It took away the “status” that comes from being a founder in a more burgeoning scene. You’re not cool for being a founder in SF, everyone is a founder. To be a cool here, you need to be really, really good at it.

While I love San Francisco, I don’t think there are other geographical reasons to be here. If you picked up the top 5,000 founders from the bay area and dropped them in Iowa City, I suspect that would become a more compelling place to build a startup.

I learned about gradient descent

Going from an initial startup idea to PMF is an iterative process not unlike gradient descent.

Let’s say your startup is some N-dimensional vector, with every value in the vector representing something about your thesis - the specific problem you solve, who you solve it for, how you solve it etc. Your growth rate (perhaps the most important metric) is a function of this vector. Crudely, job is to optimize this vector to maximize your growth rate.

This is one of the reasons for the canonical advice of being extremely specific about what your startup does. You should be precise to begin with and then search for accuracy. You need a specific vector so you have something to optimize.

“We are building software for dental offices” is not specific enough to optimize, there just aren't enough dimensions in this vector.

“We are building a tool that receives phone calls from patients requesting appointments or reschedules and automatically updates calendars for small dental practices in the US” is pretty specific, and there is a lot in there to optimize.

Miscellaneous Learnings

1. Don’t go to “founder events”. Real founders aren’t there, they are busy working.

2. Founders sometimes go to industry or tech specific events. You might see good founders at “ETL and drinks meetup” but unlikely at “founders and VCs mixer”.

3. Don’t talk to investors until you have at least some traction. It’s a distraction, you should be talking to customers.

4. When you do want to raise money, talk to a lot of investors all at once. Time-box the fundraise and do it deliberately.

5. Be radically empathetic towards your customers - it’s your advantage. Worry about how you will scale this empathy later.

6. Charge your customers NOW. Not tomorrow, not when your product is “ready” (Hint: it never will be). We charged $5/mo for unlimited Greptile usage when we started. This was a mere 8 days after first line of code. Any non-zero number is fine. 0 is bad because even if you do get users, you’re learning nothing from them about whether or not they want the product until you ask them to pull out their credit cards.

7. When people ask you what your startup does - answer the question in the first sentence, and then provide details if they are still listening. Don’t bury the lead.

8. Try to find a co-founder, and ideally make it the smartest, highest integrity person you know. There were countless occassions where without the support of Soohoon and Vaishant, a tough month would have knocked me out. Conversely, the wins are always 10x more rewarding.

9. Don’t worry about competitors - odds are you will all die, best focus on yourself.

10. The exception is competitors that are huge incumbents that customers like. You should probably believe something to be true that they don’t.

To conclude - Silicon valley is small - be kind and helpful whenever you can. It will come back to you fast. I like to think I have positively impacted many people I have met over the past year, from making customer intros, to hiring intros, to YC referrals. Much more than that, any success Greptile has had in the last year has been a product of the kindness we were offered be complere strangers. Its a beautiful thing.