Monthly Archives: June 2016

Romulus Capital

My first experience as a tech startup investor was with Romulus Capital. We founded Romulus together with Krishna Gupta and Anant Chaturvedi in 2008, while still students in college. Krishna was 20 years old, and I and Anant were 21 at the time. We were young and new to the game and this was reflected in the size of Romulus’ first fund which was less than $1M.

Our original plan was to invest $25K-$100K checks as the first investor in companies emerging from the MIT campus. Seed rounds were much smaller in 2008 than they are now.

We later expanded to invest in companies across the East coast as well as some on the West coast, and achieved two exists when Crocodoc was acquired by Box and Gyft was acquired by First Data.

I left Romulus after the first fund for my current role, and Anant is now working at his family business Uflex, but Krishna kept going. Together with his new partner Neil Chheda, they raised a $50M second fund in 2014. And earlier this month they announced a $75M third fund.

As you can guess, the vast majority of people who we pitched the first fund to while we were still college students thought we were crazy. While we were launching a fund rather than a tech startup, it was a startup fund. As a result, we experienced as many rejections for our funding requests as most startups do.

In light of these experiences, it’s amazing to see Romulus’ current success. I congratulate Krishna and Neil on what is the latest but certainly not the last step of their journey.

Artificial intelligence and deep learning primer

Artificial intelligence is a hot topic these days. Computer programs are beating humans at complex games like Go, cars are learning to drive themselves, and more and more startups are claiming that they use artificial intelligence in their products. With this much hype, it’s easy to overlook the fundamentals of the subject.

What is artificial intelligence, what are its potential use cases, what does its history look like, what’s the difference between artificial intelligence, machine learning, and deep learning, and what is the current state of affairs in the space?

Here’s a great primer by Andreessen Horowitz which answers these questions.

Hande from Insider

Insider‘s co-founder Hande Cilingir recently gave a talk about web and mobile real-time personalization at the ArabNet conference in Beirut.

In her talk, Hande defines real-time personalization, shares global and regional e-commerce statistics to show why it matters, and gives examples of how Insider’s customers use the company’s software to deliver personalized user experiences.

You can watch the full talk below.

Tapu opens to individual sellers

I wrote about our investment in online real estate auction marketplace Tapu in a post from August 2015.

At the time of Tapu’s launch, it sourced the supply of properties auctioned on its marketplace from banks and asset management companies. These organizations had a stock of properties that they found challenging to sell through their existing channels and they welcomed Tapu as a new sales channel. The properties were also a good fit for Tapu because they came together with land titles and valuation reports which gave potential buyers the confidence to bid for them in an online setting.

However, Tapu soon began to receive requests from individual sellers who also wanted to auction their properties on the marketplace. Individual sellers represent a larger supply of properties than banks and asset management companies, so this was a big opportunity for Tapu. The problem is that, unlike the properties of banks and asset management companies, the properties of individual suppliers don’t always have land titles and rarely have valuation reports. This makes them difficult to sell in an online auction environment.

Rather than accept these problems, Tapu decided to solve them. And the result is that Tapu now lets individual sellers auction their properties on its marketplace. To do so, Tapu connects its individual seller candidates with licensed valuation companies who confirm the land title and provide a valuation report for the individual sellers’ properties. This verification and appraisal process ensures that potential buyers can have as much confidence bidding for the properties auctioned by individual sellers as they have bidding for those auctioned by banks and asset management companies.

The result is a new and efficient sales channel for individual sellers who don’t have to deal with the back-and-forth communications, property visits, and price negotiations which are part of the experience of selling a property through a classifieds site. And the result for buyers is that they have many more properties to choose from.

The development is a win-win for both sides of the marketplace and I believe that it’s going to be an important source of growth for Tapu.

Healthcare problems

I visited the hospital earlier this week. I injured my fingers after tripping and falling over while running outside, and two of my fingers were in pain and swollen following the accident. Rather than wait it out, I decided to see the doctor.

The doctor identified that one of my fingers was broken and recommended that I undergo an operation to fix the fracture. Without going into specific probabilities, he shared that there was a good chance that the fracture would heal itself, but that he was recommending an operation to be on the safe side. He also shared that, if I decided to undergo the operation, it would need to be done within a week as it’s easier to operate on a broken finger than a partially healed one.

After hearing this information, I had two questions. What’s the doctor’s track record in correctly diagnosing the need for and correctly performing such operations, and what’s the price of the operation?

In most sectors, you know what you’re getting and what you’re giving. A product has consumer reviews and a price. This isn’t the case in healthcare.

Although there are doctor review sites, including our investment Doktorsitesi, the reviews at these sites only share what patients thought about a particular doctor. These reviews largely reflect how comfortable the doctor made the patient feel rather than the accuracy of the diagnosis and the success of the subsequent treatment.

In light of my specific circumstances, the reviews that I want to see include the fraction of a doctor’s visits where he recommended an operation (including how this figure compares to that of other doctors in the same role), and the success rate of the operations he performed. The first is a proxy for the correct diagnosis of the need for an operation (sometimes doctors recommend operations even though the patient doesn’t need them in order to earn more money), and the second is a proxy for the doctor’s abilities. Doctors and the hospitals that they work for have this information but they don’t share it with patients.

Getting information about the price of an operation is equally challenging. I asked my doctor what the price of the operation would be and he directed me to his assistant. I asked his assistant who directed me to the hospital’s administrative staff. The hospital’s administrative staff asked what percentage of the price of the operation will be covered by my health insurance. When I shared that I don’t know, they followed up with two potential prices for the operation.

The first price is if my insurance covers 100% of the operation, and the second is if I need to pay out-of-pocket. The first figure is higher than the second. But I wouldn’t be surprised if the first figure is just a headline number shown to the patient and the insurance company actually has a separate agreement with the hospital whereby it pays a lower price for the operation. I think you get the picture.

Our health is our most important asset. But when it comes to making buying decisions for our health, we don’t know what we’re getting, it’s very challenging to find out what we’re giving, and what we’re giving reflects our ability to give rather than the quality of what we’re getting. These are all problems to be solved.

Captain of a large ship with a small rudder

I recently listened to an interview with Elon Musk at the Code Conference. Elon shared his thoughts on a wide range of topics including when autonomous cars will be ready, how soon we’ll be able to start sending people to Mars, and whether we’re living in a simulation. I recommend you watch the full interview below.

However, the most interesting part of the interview for me was when he shared his view that “Being US President is like being captain of a large ship with a small rudder – there’s a limit to how much damage they can do.”

Elon shared this thought in the context of the current election cycle in a democracy like the US. This is why he focused on the damage that a potential President can do. However, the same is true for the positive impact that a President can have. For example, there’s only so much that a President can do to drive short term economic growth during his presidency. We tend to attribute much more of a country’s economic outcomes to the policies of the President in power during that time than is really the case. Factors like a country’s sectoral innovation, private and public debt levels, current account status, and monetary policy are all fully or largely outside of the President’s control.

The same is true in entrepreneurship. From the outside, it’s tempting to ascribe the successes and failures of startups to founders. It’s tempting to make heroes and villains out of entrepreneurs.

However, this is a misattribution. Just like being President, being a founder is like being captain of a large ship with a small rudder. The ship is the startup, the sailors are the team, and the waters are the market dynamics. The sailors and the waters have a greater impact on the ship’s destination than the captain. The captain’s most important decisions therefore consist of picking sailors, deciding which waters to navigate and how, and keeping sailors motivated during the journey. That’s why recruitment, setting vision and strategy, and establishing culture are a founder’s most important responsibilities.

The value chain

I was recently speaking with one of our marketplace entrepreneurs. They currently serve the supply which they receive from supply aggregators to end customers. So they have a direct relationship with end customers but only an indirect relationship with the individual sources of supply who work instead with supply aggregators.

A recent industry development gave the company an opportunity to start working directly with the individual sources of supply. By incurring the team and software development costs necessary to source directly from individual suppliers, the company can effectively become an aggregator itself, thereby avoiding the commission that it currently pays aggregators. The question, however, is whether this makes sense.

Serving as an aggregator moves the company from one part of the industry’s value chain (that which serves the end customer) to another part (that which is one step removed from serving the end customer). And depending on the industry dynamics (for example the competitive set of companies serving end customers, the competitive set of supply aggregators, and the negotiating power between supply aggregators and companies serving end customers), players at different parts of the value chain have different abilities to capture the value created by the industry. Supply aggregators may capture more value under certain industry dynamics and companies serving end customers may capture more value under different dynamics.

Before forwards or backwards integrating into a new part of your industry’s value chain, it’s important to evaluate just how much value there is to be gained from making such a move.

Investing as a generalist

I was recently speaking with an investor in the US who asked which sectors we invest in in Turkey. I often get similar questions about what business models or stages we invest in.

In a market like the US where many successful tech startups are founded each year and there’s a lot of competition to fund these startups, many investors specialize in specific sectors (like fintech or biotech), business models (like e-commerce or marketplaces), or stages (like seed and early stage). This gives them a competitive edge in evaluating startups in their target subset and lets them differentiate themselves as a source of capital.

While this approach makes sense in deep markets like the US, I don’t think it works in a more shallow market like Turkey. The reason is that there are few tech companies that go on to reach the $100M+ valuations necessary for a venture capital model to work in Turkey.

By my calculations, 15-20 tech companies that eventually reach a $100M+ valuation have been founded in Turkey over the last 15 years. I may have missed some companies but the number is in that ballpark. That’s 5-7 such companies over a fund’s traditional 5 year investment window. Assuming a fund makes 10-20 investments during this time, that’s enough to pick from only if you’re sector, business model, and stage agnostic. If you only focus on a certain subset of companies across one or more of these dimensions, you make it much less likely that you’ll get into enough winners to build a successful fund.

As a result, we take a generalist approach to our investments in Turkey. We invest in the transport sector, but also in the fashion, hospitality, finance, and real estate sectors. We invest in e-commerce, but also in marketplaces, SaaS companies, and classifieds sites. We invest in seed and early stage rounds.

As a result of our generalist investment strategy, we’re reactive to the ideas and market approaches put forth by startup teams. We invest in teams that we believe have found or will find the right strategy in a given market rather than try to proactively determine the winning strategy for a market ourselves and then look for teams taking that approach.

TED talks parody

TED talks are short talks where thought leaders on a particular issue share the research and ideas which produce an often unconventional insight about that issue. Although the talks were originally in the areas of technology, entertainment, and design, they’ve since expanded to include talks in pretty much every area imaginable. I really enjoy the talks and recommend that you check out a few on the TED website.

However, although the insights which result from each talk may be unconventional, the structure of each talk is quite conventional. And this predictable structure means that TED talks can be parodied.

For this parody to be meaningful, I first recommend that you watch a few actual TED videos if you haven’t done so already. Then watch the parody below.