Intersect LabsAnkit Gordhandas | 2020-04-30
Could you give us a quick introduction to Intersect Labs?
Intersect Labs was founded in late 2018. The idea was that a lot of companies and teams out there have a lot of historical data, but they don’t know how to convert this data into predictions unless they are one of the few teams in one of the top few companies. They can’t make that conversion simply because data scientists are rare and expensive. That’s why we were founded. We were founded with the idea that anyone who has access to historical data, even on an excel spreadsheet, should be able to make accurate predictions about their business.
In terms of our funding, we were part of the summer 2019 batch of YC (Y Combinator). Until then, we were bootstrapped. Right after YC, we raised some money for our seed round.
What was your career journey prior to starting Intersect Labs? How did you come across the idea?
I graduated from MIT with a degree in Electrical Engineering and Computer Science ten years ago (2010). I was actually looking forward to our ten-year reunion but that got cancelled because of the current situation around Coronavirus. Since graduating, I have been doing a bunch of signal processing, data analysis and machine learning type work. I started in academia, starting off at MIT itself and then working for Massachusetts General Hospital.
Following that, I transitioned into a role at a medical device startup and then worked at a tech startup that made garments that measure muscle activity. After that, I had to move from the San Francisco area to San Diego, because of my wife’s career. I decided to take a little bit of a break. For a couple of months, I worked at a company called Gyroscope. Then, I started consulting for a few companies that needed machine learning help. I was their first machine learning engineer or their first data scientist. That’s when I realised that a lot of these companies wanted to do these things themselves, but there were no tools out there to support them. That’s where the idea for Intersect Labs came from. Could I build something that would empower these folks to start doing a lot of these tasks themselves?
Were there any other ideas that you were considering when you were starting out?
Not really. Not in earnest for sure. I obviously did want to start something so every day there were a hundred ideas going through my head, but none of them were nearly as serious as this one.
How did your previous experience prepare you for your current venture? In what ways were you unprepared?
From a technical perspective, I was fairly well prepared but even then, I was not completely prepared. I’d done data science and machine learning before, but I’d never written a software product end to end. So that was tough for me – it brought lots of interesting challenges. I got plenty of help from friends. One of my friends actually helped us build the first version of the product. He couldn’t join at the time, but a year later he became our first employee.
I knew from the get-go that I didn’t have a sales background and that I’d never done any sales. That was a big question mark for me. Also, the logistics of running a company can be daunting. Fortunately, right around the time I was starting this, I went through Y Combinator’s startup school. That helped a lot. It was very good in the sense that we had a small group of peers, we were assigned a mentor who would do a check-in every day, and there were a bunch of lectures. It was all about getting through to potential customers and selling your product. When I say selling, I don’t necessarily mean selling in exchange for money. It could be selling in exchange for feedback or giving customers some access in exchange for letting you observe them. The idea is: launch your product fast, launch in front of users and learn from that.
After you had the idea what were the next steps? How did you recruit your first user?
The first step was to figure out whether this was technically feasible. That took quite some time, I’d say about three to five weeks before the first prototype was built. From there, I reached out to a bunch of people I knew and asked them to try it out or give me feedback. It started with friends and family. None of my friends and family were in a position to buy the product, but I got their feedback.
Once I started increasing that circle of people I was reaching out to, people I’d met at a conference or friends or friends, that’s when we had our first a-ha moment. Our first positive response was from a friend of a friend who I’d met at a conference. That person, his name is Adeel, at the time he was CEO of a YC startup called Medumo, which has since been acquired. I reached out to him to ask if he would take a look or if he knew anyone who would be interested in the product. He responded, either the same day or the next day, saying that they were looking for a solution like this. That was really a magical moment for us. Until then, we were trying to get people to take a look at the product or try it out. Now, we had validated that there is at least one company that needed this. At the time, even having one company say “we need this” rather than “this sounds cool” was pretty magical.
Were the friends and family that you reached out to in the corporate sector? Or did you spread your net wider at the start?
Wider at the start. One of my calls was with my parents. They’re doctors, they’re never going to use this, but I was still getting feedback. And I was just planting the seed in their heads. Because who knows who they talk to? Initially, it was from the perspective of “I want feedback”, not from the perspective of “I want to sell you something”. People are more willing to talk to you if you’re not trying to sell something.
Did the idea evolve as things progressed?
The core idea still remains the same. We want to make predictive analytics easy for the non-technical user. Over time, as we’ve talked to users, we’ve added a lot of firepower to our product. For example, one of the big blockers we discovered early on was that a lot of companies that wanted to use machine learning had data, but the data had to be restructured. In the spirit of early startup founders, I would just offer to take a look at their data and restructure it. Over time, we realised that this was a repeated problem. So, we started building this product called Pasteur. We launched Pasteur a month ago. The idea is extremely simple but it is an amazingly powerful product. It lets you import your data and build a pipeline that lets you define a series of transformations that you want your data to undergo. It’s all drag and drop; there’s no coding required. Once that pipeline is built, you can keep re-using that pipeline on different datasets.
Another thing we discovered was that people like to use machine learning to plan for what-if scenarios. Let’s say you are a company that sells a lot of products of Shopify and you’re interested in finding out how changing your marketing spend is going to affect the number of goods sold. Early on, when we discovered this, to validate the idea we built a simple google sheets plugin. Since then, we’ve gained traction on that product and have added a simulator to our platform.
We constantly keep adding things to our platforms. The best way to decide what to build is by talking to users. Every single thing that has been added to our platform has come about because users have asked for it or we have observed users needing it. That is great because once you have built something, you know for a fact that people will use it.
What was your experience of Y Combinator? Any tips for readers on getting into a program like this?
I think the experience of going through YC was fantastic. It’s magical when you put a bunch of dedicated, committed, individuals together in one place. And then you add a bunch of people who have seen companies fail and succeed over the last few years. It makes a world of difference. During YC I learned about talking to customers, shipping features, reaching out to potential customers, how to price things. Because we had mentors or partners at YC who have seen this so many times before, we had really good advice. I had an office hour with the person who was the Product Lead on the Growth team at Airbnb yesterday. It’s amazing how these people are literally at your fingertips; I could WhatsApp them any time. So that was great. The added benefit of YC was that fundraising became a lot easier, but I would say that’s a by-product of being able to build a really good company.
I think the tips for getting into YC are not really orthogonal to tips for building a successful business. That’s just talking to users regularly, trying to build things and launch them as fast as possible. The biggest predictor of startup success, according to the head of admissions at YC, is speed. Any accelerator out there, I imagine, are looking for teams that move really fast, can adapt to customer feedback quickly, and teams that are disciplined. You don’t want to make random decisions; you want to make decisions that are driven by logic and data. Honestly, that’s it. If you think of something, build it, launch it, iterate on it. If you do those things you will be on your way to building a successful business and there is a high likelihood you will get into one of these accelerator programs.
Any funny or most memorable stories along the way?
The most memorable moment involved a fairly famous company, which I won’t name, within the YC network. I was really trying to get the CEO's attention so we could get them to use our product. I emailed him a couple of times; he was super busy. He responded a week later saying “hey, sure we have data and we’d love to make predictions but honestly, our data is not organised very well. So, you’ll have to wait a few months before you can use it”. I feel like many other people would have just circled back a couple of months later. Instead, I decided to take my shot and said “How about I help you organise that data? Because I’m also good at that and can do it quickly”. The next morning, I looked at my twitter feed and the person had paraphrased our conversation and said: “I love founders who offer to help in areas which are in their area of expertise but not in the area of expertise of the company, just to help us out”. That was awesome. He invited me over for lunch and gave me a lot of good tips, so that was a lot of fun. The fact that the company decided to use our product was the cherry on top of the cake.
If you could go back and give yourself one piece of advice, what would it be?
I’d give myself two pieces of advice, actually. One is, be resilient. And the other is, focus on users. I feel like those two things, from what I’ve seen from other companies that are a few years ahead of us, are the two best predictors of startup success.
What are your favourite books?
I think there’s two. One that I read in 6th grade, How to Win Friends and Influence People – Dale Carnegie. A lot of these self-help books end up being fluff. But I was impressed by that book and I continue to use learnings from that book, even today. It’s definitely helped me build stronger relationships with people. At the end of the day, that’s what matters – business or no business – just having strong relationships with the people around you.
Another book, that I read more recently, is Bad Blood – John Carreyrou. It’s about Theranos. From a storytelling perspective, it's amazing. It reads like a novel even though its non-fiction. But if you dig deeper into it there are a lot of lessons there for young entrepreneurs. As a founder, you need to understand where to draw your own boundaries, and that book is a great case study on how not to draw your boundaries.
Learn more about Intersect Labs at: https://www.intersectlabs.io/
Learn more about Intersect Labs at: https://www.intersectlabs.io/