COO Oyper: “Our technical solution finds clothes from video most accurately”

Co-founder of big-name startups Oyper and Oybox told what’s wrong with Belarusian investors, how to structure a pitch to get funds, and how his projects will take the world.

Belarusian startup Oyper has been loudly announcing itself during the last year, as it should. This activity attracts customers, investors, and partners. One of them became a co-founder of the second startup named Oybox. It is luxury outfitting brand, whose main employee is a neural network. We discussed these startups, industries to survive in and fundraising with Kiryl Sidarchuk, who is a co-founder and the heart of these projects.

The right background to launch a big-name startup

I’m in business for over 10 years: I had been dealing with both online and offline commerce, working with suppliers from all over the world. I have two diplomas: the first one is Secondary Education (I’m an English teacher), and the second is the Master of Economics.

First attempts to earn money and support my family were in my childhood.

My parents tell that I started talking late, and "money" was the first word I said.

I earned my first money when I was 10. I took my granny’s strawberry, sold it in Minsk, and all the profit was mine.

What is Oyper, who are the first customers, and what about rivals?

Oyper is an American company consisting of 13 teammates. Our main focus is the integration of visual search by image or video in e-commerce. Now we are focused on clothing because it is the most popular in Internet commerce. It is also the most difficult domain for computer vision if we talk about qualitative recognition, classification and search for similar ones.

Our goal is to give the most accurate solution for the user because it doesn’t matter which company has provided the service to him, how much money has been invested, how many people were involved — the main thing is the result. (Editorial comment: There are few competitors like Sloy from Yandex, Screenshop by Kim Kardashian, solutions from Amazon) And we do not see rivals for our solution: we have a specific architecture of a neural network, our own data set and a dictionary for marking it up. This cannot be re-created by someone else.

2.5 years ago, we were going to create an AI stylist, build a neural network that could make completed looks from items available in a catalog, self-study, take into account the actions and features of a user. The first step was to build a neural network that can recognize items of clothing in a picture with most accurately. It’s not enough to know that trousers are trousers, there are a huge number of details: style, cut, materials. So the neural network was the first thing we did. At the very beginning, our plans were different from the existing product, which has now become the main one.

We set the highest bar in the industry and went straight to the largest suppliers of video content. It were regional telecommunication companies with millions of subscribers.

We now have 3 letters of intent on our technology supply.

But we have insufficient resources to serve all these clients, now we can work with only one.

Early crisis and decision to move on

At the very start of the project, we had a team crisis when the co-founder, who was responsible for artificial intelligence, left. It was a heavy blow for us, because we have an AI-startup, AI-development, and it just stopped in one day. He left because he did not agree with our main course. The rest of the team members had a common vision, and he had his own.

We could not find someone to replace him for a long time, because he is really an entrepreneur and a co-founder, and he behaved accordingly. He was ready to spend nights researching if something was wrong, he didn’t give up, didn’t say “give me a way to resolve it” — he looked the way out himself. That is, it really was a huge loss, the most serious crisis in 2.5 years.

We decided to move on. Finally, we found an employee to replace the co-founder which had left, and now we do not concern that one of us can stop everything. We learned a lot of lessons and built a plan that we are following up to this day. We applied the best practices of American law like a cliff and vesting (Editorial comment: a startup participant receives 25% of his share (vesting) after a threshold of 1 year of work (cliff); the rest is added monthly). This co-founder renounced his share, it belongs to the company.

Barcelona startup competition winner benefits

In May of this year, Oyper won a startup competition in Barcelona. The event itself was very important. But I would say that it was the masterful work of the team that added the best value. And most importantly, the skills of the person who frames our marketing and PR strategies. We got over 20 applications for our demo from different formats of companies for the first week; work has begun more than ever before.

The main advantage of our team is that we can get the most out of every dollar. Everyone has this key skill, and we all use it.

Now we want to separate move away from the title of a startup. We want to be a stable, substantial business in the eyes of our clients. We create the image of a company you can trust to, not a startup that will practice on you. This is important because the video providers let us into their players and retailers into directories.

Oyper got Google credit. Is it possible for other Belarusian startups to do the same?

The Google credit limit has already exceeded $100 thousand. Our CEO works at WeWork in New York, and he has access to the network, which enabled him to make contact with Google Cloud representatives. He talked about us, present the pitch, they watched our demo, looked at what we were doing, and offered us $20 thousand. It was last year. We started using these credits, and they watched what we do with them.

Six months later Google offered us to become trusted tester partners.

In January 2019, they rolled out a test platform for people like us. There are a lot of opportunities: you can place data sets, train neural networks. We were one of the first in the world to receive this. They looked at what we are doing with this, they looked at our progress, and a month ago they approved $80 thousand of credit for this year. That is, we have everything cloudy on Google servers, in the same place we train our models.

To emulate this success any Belarusian startup needs to specifically talk about itself, correctly formulate the concept of what it does.

I can only talk about AI: if someone has a really worthwhile project, has a prototype, demo or idea, then he needs to carry it to customers right now, carry it to Google, carry it to everyone to get feedback. Here is a lifehack: do not to waste time developing a product until you receive feedback. Make a presentation and go ask people what they think about it. But you need to understand who is to be asked for advice. Don’t ask an offline Belarusian businessman or investor what he thinks about your AI startup, which plans to revolutionize the world.

Belarusian investments features

The truth is that there is no competition among investors in the Belarusian market. If it was, they would move faster, make more transactions, they would quickly realize what is going on.

Investors ignore those who are worth investing. Belarusian investors are open-minded and highly educated, they understand everything about everything, know the specifics of all industries, so they are a tough audience. On the one hand, they expect your hundredfold growth in six months, and on the other, they say, “Well, we need time to look at you”.

People talk about unicorns, but they never saw them. People want to invest in them, but they are not even close to them and do not know such processes.

How does this correlate with reality and the investment market?

Here is a striking example. Three days before the Barcelona R&B Next Awards Disruptive Tech contest, we had pitched in Belarus. And we were told that we are deceiving ourselves and people, that we created useless things, etc. And in a day of the competition in Barcelona, there were CTOs of Ikea and other large e-commerce organizations, we won. This situation is one of the reasons why we abandoned the idea of searching for an investor in Belarus.

To date, we have not found the person we are ready to work with. Sometimes it’s better to get by with your resources and work harder than take a person to your team who, for his money, will simply break all processes. Our approach to fundraising is very specific, we are not overstrained in the search for investments. If there are none, it will not slow down our process much. And if it is, okay, this will speed up our process, but I clearly understand who we can borrow money from, and we have not met this person yet.

Who is an investor and how to prepare for a pitch

You have to understand that an investor is a businessman. Someone sees them as people from another planet, someone considers them stupid or short-sighted. A professional investor is a well-mannered person — he will always smile and nod, and it will seem to a new person in this industry that he liked everything, but he smiles and leaves then. I know how to build a pitch logic so that a person immediately understands whether he should spend time on me and trust me with money, or not.

The venture investment market is not an offline business where a person wants to see the growth of + 10% or + 20% per month.

The general picture matters: what are our plans for the next 5 years, how we are going to achieve this, and a more or less clear plan.

It is necessary to take into account the specifics of each investor — I always prepare for a pitch, try to find out what kind of person he is, what he was doing, what kind of background he has, and for that, I adjust the pitch. I try to spend my time only on someone about whom I know right away that he is interested.

Oybox: a neural network as a replacement for a stylist and 130 orders in the first month of work

Oybox is a box with 5 pieces of clothes, and you can combine it in 6 different looks. This box is the result of our stylist professionalism and neural network.

We have more than 130 orders in the first month after the launch.

We had started in July, then summarized the statistics — the conversion from order to buyback is more than 70%. Moreover, we have 4000 unique visitors to the site and 7000 organic traffic.

I believe that for the first month of business in Belarus this is a good indicator. The first customers are already returning, although according to our calculations, our audience will buy 3 Oybox boxes per year.

We got funds in less than a week. We needed a small sum, let’s call it the first stage of the pre-seed round.

We got it on Skype, 7 minutes were enough for us to have money next week.

What was the point of this conversation? We talked about what we want to do in its global sense. Our monetization model is very complex, our plans are difficult to understand. But we clearly know how to implement them. We brought it all to the pitch, and the man fell in love immediately with the project, like everyone who heard about it do.

The future of fashion e-commerce

I am sure that those shopping methods that do not cause additional trouble, will have an advantage. People want to see, tap and immediately have it in the shopping bag. Our task is to create this seamless experience.

We have been researching the fashion e-commerce market for a long time: 66% of purchases in the USA are made impulsively.

A person sees something that he likes, he is inspired by someone in a social network, on the street, in a video — if he doesn’t make this purchase right now, the impulse passes. And we give the viewer a shopping basket, we turn his remote into a shopping tool. We believe that this is the future. And you can buy not only clothes — anything is available for purchasing.

Subscribe to the project news

One message per week, no spam!