do not cut the product for years “
“I’m half developer, half entrepreneur. If I can build something fast and effective, I do it myself, ”Nikita Kolmogorov presents himself on his website. In the summer of 2021, he sold anti-spam bot for $ 329,000. The buyer was the 1inch.io crypto project. At the time of the deal, the bot had 20,703,474 users in 190,621 chats. At the same time, Kolmogorov is the author almost three dozen other projects. At the request of Megaplan, he spoke about the features of creating products for Telegram and gave some advice to aspiring entrepreneurs.
Shieldy bot
The idea was born when my own active and popular Telegram community began to receive a lot of spam. At first I gave out admin panels to active participants, but then I felt sorry for their time. All spam was of the same type, and bans could be automated.
I wrote the first version of Shieldy in two days. This was in 2018. Newbies in the chat either went through a small captcha, or were sent to the ban. Then for several years I added one new feature after another. The main part remained simple: I added a bot to the chat, gave it an admin panel – I forgot about spam.
There are no new solutions here, I saw this in old web chats back in 2007.
Shieldy’s obvious competitors are Fyodor Skuratov’s ComBot, Oleg’s Controller Bot, and recently the Miss Rosa bot. For some reason, competitors are trying to do all-all-all in one bot. I have a slightly different philosophy: each product has one goal-function.
At the time of their launch, Telegram had the most convenient API for developers. Facebook Messenger is limited, groups are rarely used there, and social growth of the bot is difficult to ensure. Other messengers like Viber and WhatsApp have a bad API, I can’t work with them.
Today I have 25 other projects. I want to reach a billion users, now there are over 60 million.
Most of my products are free, so it’s not that important for making money whether users stay or leave. A commercial product works for the audience that uses it. I don’t need this, so I can afford not to be led by the wishes of the users.
Ideas, benefits and investors
Ideas for creating bots arise in the same way as for other products. I see the “pain” – I write it down in a long list of ideas, which then, maybe, I will implement. There are even the most insane, unnecessary, stupid or unrealizable thoughts on the list.
At some point “inventing muscle“ develops enough to instantly find solutions to any pain you notice. And when I have free time, I open the list, choose the simplest or most interesting problem and solve it.
There is no need to look for investors or partners who are ready to purchase the development. Create products that benefit people. If there is a benefit, there will be demand, and the rest will come. The key is to empower users to benefit.
You need to promote a bot like any other product: talk about it everywhere and everywhere. I have postyourstartup.com, and I use this list to post every product I launch. For more information on how to promote a bot, I wrote…
Features of bots in the “cart”
One of the problems of bots in Telegram: they cannot write to users first. Therefore, the captcha cannot be passed in PM – only in the general chat. The bot can write to the user a certain number of messages per hour.
Bot API – this is what we see literally at bot.telegram.org/api. This is a description of the main cases and methods of working with bots in chat.
User API are used by all users – telegram clients. You can also make bots through it. But there are no convenient TimeScript or JavaScript frameworks that I use.
Telegram’s bot support does not always work quickly and efficiently. There are a number of hardware limitations of the Telegram API (as far as I understand, there are two teams working on the Telegram API and the Telegram Bot API).
Every time I create popular bots, I stumble upon Telegram restrictions. At one time, Shieldy made requests to Telegram too often – because the bot had become very popular – and this was causing the messenger to malfunction. The company makes billions of dollars, and I – small, with $ 20 servers – bring their servers to error 504.
I create developers, not look for them
I use TypeScript everywhere. This is a universal development language, I can write everywhere in one language, I control the whole process. If I open a large project in Go now, I will be tearing my hair out. The code on it is difficult to read due to abstractions and limitations such as no exceptions.
The Shieldy code is publicly available, anyone can look at the technology stack. These are Node.js, TypeScript, Telegraf.js, Mongo.
I’ve been developing software since 2011-2012 and don’t like to write code (although I can do it well). Every time I manage to write less code, I am happy. Therefore, I am constantly looking for frameworks that allow almost no writing. I use this particular stack of technologies to satisfy my laziness.
Plus, I can always find developers who will be writing code from day one in TypeScript. They don’t need to understand a large Ruby project, for example.
When I need a new developer to join my team, I can find a javascriptor. There are many of them on the market, and their work is inexpensive. I’ll teach it in a couple of days, and in two weeks they will write the same clean code as me. Therefore, I do not use, for example, rare Haskell on the server. I can write functional code on it, but I can’t find a developer for reasonable money who could support it after me.
My bot Temply implements 4 steps: take javascript courses, take courses on NodeJS, understand TypeScript, write specific code in a specific repository. These steps I give out gradually: one step is finished – I give the next.
The training takes from 2 to 16 weeks. After him, newcomers can get a job as a junior in some office, but they are not hired without experience. And I offer an unpaid internship with me.
All my products remain in production forever, these developers can add them to their portfolio if they have worked on them.
During two years of teaching at a local bootcamp in Vancouver, I learned how to quickly grow good developers from beginners and found myself the best talent. On other tech stacks, this would not be possible.
Costs and investments
I invested $ 350 in the registration of a company in Canada. It’s called Borodutch Labs Inc. Through her I pay for the work of the contractors.
I co-founded another company. Samsung, Slow Ventures recently invested in it (they invested in CoinBase, Robin Hood, Citizen) $ 655,000. This company is developing the pseudonymous social network Dosu. There is no personal attachment, but you can use the social capital of other sites.
Speaking of costs, I have a pool of designers that I partner with on an hourly basis. It’s the same with programmers: most of them were newbies, but after a couple of months of working with me, they learned how to write middle-developer-level code.
It costs about $ 2,500 a month to pay their wages in the main company. Compared to the market, this is comparable to the salaries of one senior or middle + in Europe or the CIS. And I have four employees, some of them part-time. I try to keep everyone working no more than 40 hours a week. Everyone should have a private life.
I spend about $ 250 a month on servers, for all projects. This is a penny. The company, of course, makes much more money.
Shieldy was hosted on DigitalOcean on a $ 20 droplet prior to the 1inch purchase. Where is it hosted now and on what server, it is better to ask the buyer’s team. Scaling is easy too: Node.js and Mongo have great built-in tools for both vertical and horizontal scaling. You can read the documentation for about 20 minutes – and quickly deploy everything.
Toxicity in Russia and criticism in the global market
The first rule of doing IT business is: don’t do it in Russia. The Russian-speaking audience does not like to pay, but very much loves to hate. If you are not going to cheat and are building an honest business – translate the product into English and start from foreign markets.
When I posted something on VC, I came across a lot of hate. I learned to use this anger, not to take it to heart. I answered every comment, so haters only increased the popularity of the texts.
Let’s take a look at Reddit: I have critical comments there, but they are not toxic. A person can calmly explain where there are mistakes. He doesn’t say, “Ha, I got lost, you have a mistake, ahaha.”
If you want to start in the Russian community, please. But be sure to translate the bot into at least English. Voicy partly became popular because it was translated into 20-25 languages at the start. The first 50,000 users came from Iran. For understanding: there was no Persian in the bot, because one of the free engines turned it off. Voicy was originally Persian, which brought the first 50,000 users from Iran. About a year later, in 2018, the free speech recognition engine turned off Persian. Voicy already has 12.6M + users in 2021.
The main advice: do not cut the food for months or years, but start it in 1-2 weeks. Too many people write something gigantic and complex, put it on the market, and nobody needs it.
You need the benefits that the product will bring. Small benefit to a large number of people, or a huge benefit to a small number of people.
