Translated material
AI investments from the creator of Revolut. How Ilya Kondrashov builds a smart venture fund with a nickname of the parties
This material (information) was produced and distributed by foreign agent Elizabeth Nikolaevna Ossetian or concerns the activities of the foreign agent Elizabeth Nikolaevna Ossetian. 18+
This material (information) has been produced and widespread by a foreign agent The Bell or applies to the activities of a foreign agent The Bell. 18+
Quantumlight is a venture fund that works on the basis of AI and selects startups using artificial intelligence technologies. The idea of the Fund was invented two years ago by Nikolai Gorodorsk-the founder of Fintech Startap Revolut and Ilya Kondrashov-an entrepreneur and a banker. Both of them wanted to change the traditional approach to venture investments, which now, according to our hero, are somewhere in the field of mysticism. Together with other investors, they invested about $ 200 million in the project.
We met with Ilya Kondrashov in London in the Revolut office and learned how the AI model is arranged for venture investments, on the basis of which artificial intelligence analyzes information about companies, how the final solution is made, as well as what ideas you need to come to investors, why it is better not to raise rounds in the early stages and why investors are not unconditioned for Revolut.
Chapter 1. Fund based on artificial intelligence
- Tell us how you ended up in Revolut? Are you familiar with Nikolai Sorvyskoye?
- Yes, I have been with a partial sign for 6 years. We are both fintech entrepreneurs. In 2011, together with university comrades, I founded a company called Marketfinance today. This is a credit business, where, like in any other, there were many ups and falls.
In this field, I met a part -time. He then raised, in my opinion, a round for Revolut to estimate $ 40 million, which I tried to bind. But he said that, unfortunately, he could not let me into this round. I did not particularly take him [pursue. - approx. The Bell], but it was necessary.
- He chose who to take the money from whom to take?
- Yes. But I failed to get into this round. Then he tried to hire me in Revolut to lead the credit department here, but this directly competed with my project.
As a result, we came to the idea: why not hire the coolest programmers and create a system that would predict which startups will be successful. And Nikolai will bring his own methodology for building a business to the project. Still, Revolut was built quite differently compared to most companies, and this is a very great merit. He can give this know-how to other entrepreneurs.
- You and Nicholas, as I understand it, made a venture fund. We rarely actually talk about venture funds, but in your case I was hooked wildly that it was built, in any case declares, with the help of AI.
- Yes.
- That is, artificial intelligence is laid there. And everything that is now connected with artificial intelligence is, as you know, Hot. Therefore, for us and for our viewers, this is also Hot, and you must definitely talk about this fund. I understand that Nikolai came up with him and called you?
-I, too, hope, [brought] some part. We have long discussed this with him, about 6 years ago.
- That is, he did not take the money, but you continued to communicate?
- Yes. Once, on the Found Forum forum in London, this is a pretty good event for fauners-we met, I suggested an idea, why not hire the best programmers.
- So this is your idea?
- I can't say how ...
-Let's honestly.
-Yes, I had the idea to hire very cool programmers and build some kind of system. We decided that the venture market is ineffective. Why? Because there are some kind of patterns ... I don’t know how patterns are in Russian.
- Samples, templates.
-Not even templates, but more patterns that are stable in some long period of time. Roughly speaking, phante with higher education will be better to perform than faunters without higher education. Companies that have successful investors will be better for companies that have unsuccessful investors.
- That is, you took cases of successful startups, dismantled them for components and appropriated weight.
- Yes.
-And they laid the data in some kind of program?
- Yes, machine training.
- And she is counting on?
- Yes, the probability of success.
- Are there any results?
- Yes, we have already invested 20% of our fund. These are very promising companies in the USA, Europe. We look at India, Israel, Latin America.
- Tell us more about the fund. Is there, firstly, size?
- Yes, $ 200 million.
- $ 200 million. Do you attract money or are you yours?
- We have both our money and third -party investors.
- Own are Revolut Faunders, that is, Peninonsky, you?
- I and Setonsky. Mostly Setonsky.
- And you attract more?
- Yes. Some of them were even guests of your program.
- Can't be called?
- People do not like when they talk about where they invest.
“You understand who I thought about first of all.” About Yuri Borisovich [Milner. - approx. The Bell].
- No comments.
- And for what time is this designed?
- As usual. In Venture, all funds are 10 years. But usually money is returned in 4-5 years, and profit returns within 7 years.
- What needs to be done to enter?
- We communicate mainly with institutional investors, with all kinds of funds. Some Family Office. That is, a rather diversified pool of investors.
- That is, these are not just small checks. What is the minimum with which you can come to you?
- At least $ 5 million now. Now we are trying to limit the number of investors.
- The less, the better?
- We look more at who can work with us in long. Because we are still not limited to one fund of $ 200 million. We have plans, our platform is scale to attract future funds, for different strategies. In the private market, quantum methods make up a very small percentage of all strategies. Even if you look at Real Estate, how decisions are made there, for me it is simply unknown. It seems to me that there is also the opportunity to come to decisions based on models.
- This engine is called Alef.
- Alef, yes.
- Why so? And what is it?
- This is a game of words. Alpha is what is called superprofit. Plus Alef is the first letter of the alphabet, number one.
- Another schedule is drawn on your site that you are overtaking the market many times.
- This is a backstest.
- What is it?
- This is a very standard method for evaluating models. After we built a model for making decisions, we begin to test it according to historical data. For example, we trained it on all the information that was Available [available] by the end of 2013, and now we look at which companies it would invest in 2014.
“You are strong in your back mind.”
- Yes. But this is how the effectiveness of the model is checked.
- What would she invest?
- For example, in Revolut, she would always invest in our bacters. This is a key test for us.
- In addition to Revolut, what would she invest?
- Many: in NewBank, Snowflake, Databricks, etc.
- How much did you drive there?
- This is Bigdat. Only the profiles in LinkedIn, in my opinion, are now about 700 million, for each of them there are 6 features.
- What is the connection between Linkedin and ...?
- Roughly speaking, we aggregate data on different sources.
- Do you aggregate people?
- Yes, including people. We look at what kind people were successful in the past, whether there are any characteristics. Today we work with about 20 different sources.
-Sources are some kind of date in Linkedin ...
- LinkedIn, Glassdoor [platform, on which employees anonymously leave reviews about their leaders and companies. - approx. The Bell], press releases. Or it may be an assessment of companies.
The process of improving the data after we bought and made it with different sources - this is where our entire profit is generated. It is difficult to work with these sources.
- They are not given or they are dirty?
- They are dirty, yes. One of the biggest problems is how much information from the future can bake into the past. If, roughly speaking, to see LinkedIn, then, for example, everyone deleted unsuccessful companies or their unsuccessful experience. Therefore, if you take LinkedIn as, that is, by today's date, it seems that everyone had only successful experience.
Therefore, we need to take information in LinkedIn on some historical date-for example, for 2016. And then there is experience there ...
- A large team of programmers is working on this?
- In the general fund now about 20 people. She [team] is not very large, but she is very good. We have the winners of the International Olympiads for Programming, Physics.
-It seems to me that press releases are also a source of bargain. Because people often represent themselves better than they are.
- Yes. There is some kind of share. It is necessary to build an algorithm in such a way as to output it.
Roughly speaking, we have two categories of companies: successful and unsuccessful. Suppose we are trying to break this cloud: on the right we have all the companies where Founder graduated from Top University, on the left - all companies where Founder graduated from not Top University. Then we count the number of successful companies. And, let's say, there are all successful companies, but here all the unsuccessful companies. Here is such an algorithm. This is done automatically. We checked 300 different categories of metrics by which successful companies from unsuccessful can be divided. Now we have somewhere 50-60, which really work and carry information.
- And really top -end fauners end with Top University?
- We even checked, for example, the Drop Out combination [departure from the university, deduction. - approx. The Bell], and it works only in combination with Top University. Therefore, if you are Drop Out, then it works ...
“Only from Stanford.”
“Yes, Harvard or Stanford.”
“Are you still making calls with the Founders, Pitch?”
- Yes, we make calls. But the question is how we specifically make a decision.
- How?
- We make a decision on the model and provided that we do not find “red flags” in the company.
- What is a "red flag"?
- It can be, for example, Frod [fraud] or there may be numbers that fall into the model and are wrong.
- If there is an inner feeling, GUT FEELING, and it tells you that you need to invest, and the model says that you don’t need, what will you do?
- We will not, the hands are tied. You see, if you were an investor in my fund, I would tell you that our strategy is to invest according to the model. Then some time passed, we, for example, lost money, come to you and say: "We lost money." You begin to understand: "Why did you lose money?" "But in fact we are 90%..."
- Here Ilya had GUT FEELING.
- Yes, we had GUT FEELING. You will be unpleasant, right?
- On the other hand, it will be unpleasant for me if the model says “yes”, and your GUT FEELING will say “no”.
-Therefore, it is necessary if the model says “yes” and we have not found some kind of fraud or something else ...
- Do you trust your experience and your Chuyka less than the car?
- Yes, here is a simple example. A lot of people left Revolut and the beginning of their businesses. In total, these companies, where natives from Revolut occupy leading positions, have now raised more than $ 2 billion investments.
- Together?
- Yes, taken together. This is a rather large category of people. If you look at their performance in Revolut and how much we, even Nikolai, would evaluate their potential after Revolut, then does not converge at all.
- Doesn't it converge?
- It does not converge.
- That is, you said: "Oh, let it be wounds up, he does his own, he will not succeed in anything, a fool."
- Yes. And they often shoot. And those who, it would seem super, top performer, suddenly does not work. It is very difficult to predict the future.
- What will happen if everyone else also begins to write these programs?
- Sooner or later, of course, this will happen. If we have good results, then, of course, you need to be prepared for this. Now the level of competition is quite low. The share of quantum investments from total less than 3%.
- Do you imagine a situation in which you rely only on your program?
- This is our goal. Complete automation.
- Will you entrust artificial intelligence complete decision -making?
- Yes. This is Pattern Recognition [image recognition]. There are people who have invested in successful companies, and they can predict the following successful [company], because they have a pattern in their heads. But a person is much worse than a car in recognizing these patterns, because the car has an ideal memory and there are no emotions. There is still no such aspect as politics, when the venture investors decide, whose share in the company or in the fund more,
Then there is no part, for example, venture investors in the committee decide. The committee also has such an aspect as politics. For example, someone has more a share in the company or in the fund, and their words are already beginning to sound more significant.
- How much is the fund?
- 2 years.
- Are there any indicators by which you can already judge?
- We invest a lot in artificial intelligence applications. What we see with artificial intelligence is not a hype. There is a large share of hype, but there are really companies that grow very quickly. We see that the sale cycle for large companies, Enterprise Sales, which usually occupied up to a year, has now been reduced to 2-3 months, which is actually an incredible difference, because CEOs of corporate are very focused on the introduction of artificial intelligence. They see this as an opportunity to improve their profitability, competitive advantage. We still have several investments in Medical Ai. For example, Red AI, the company in San Francisco, makes automatic writing reports for Radiologist, as in Russian.
- Which are the X -ray.
- Yes. How it looks. The doctor has two screens. Pictures appear on one. There is a microphone in which he says that he sees in the pictures. What he says appears on the second screen. This system reads what he said and makes a report out of him.
Doctors really need this Final Note to be written with their Tone of Voice [personal communications style. - approx. The Bell]. That is, some lotions that they historically used. Therefore, for each doctor you need to train a separate model. Imagine, there is Openai ChatGPT and there is a separate model for each doctor who trains their own data, fixes their language.
- You said for 2 years. Somewhere in 4-5 years, it becomes clear what results.
- Yes, it will be clear. We already have companies in which revenue grows very quickly after we invested.
- Do you invest in stage B?
- Yes.
- Why exactly the stage in?
- What we give to the faunters is a training manual on how to raise a company. Find the so-called Product-Market Fit [your audience and niche for successful development. - approx. The Bell], finding the topic of business itself is a rather unique task. This is not to say that "Revolut did this, so you do so too." As for scaling, this is actually building internal processes for the growth stage, the approach can be more common there. We are trying to play the game where we can win. [With the transition] from Series A on a series of half of the companies does not survive. Given that from the series in money, mainly the TOP of 5% of companies earn, then this is a large level of risk.
Now the rounds of the series B is from $ 30 million to $ 50 million each, and we invest small amounts in them.
- How much do you invest?
- From $ 4 million to $ 6 million per round.
- To decompose more?
- Yes, we calculated the optimal number of companies in the portfolio, and we got such an amount.
- What is this optimality? Why 50 companies are more optimal than 100?
-Firstly, the volume of the fund is limited. You can invest everything in one company. For example, if she shoots, you just ...
- It is clear that so much.
- There is a certain diversification. It is impossible to decompose 1000 companies, because each check will be so small that even if some of them shoot, they will not pay off a bunch of companies that will not lead to anything.
Chapter 2. The beginning of a career in Goldman Sachs and the launch of Kriya
-A little about you. You, too, went first along the beaten path of the Goldman Sachs banker?
- Yes, it was fashionable then. It seemed to me that it was hard.
“You just like it when it's hard.”
- Yes, I'm not looking for easy ways. No, it was interesting there. By the way, we worked on several Russian transactions. For example, IPO Mail.ru, there was such a significant transaction in Russia. Because it was not only a very successful company in itself, but they still had a piece of Facebook in the portfolio, in which they invested before ...
-Have you met Milner?
- Yes, it was lucky for us, yes.
- What then pushed you out of there into the thorny winding path of your own business? This is a completely different, much more stressful story. It is one thing - you are a career banker, and another - you build the company itself.
- Don't know. Stupidity, arrogance, of course, if then I knew what I know now about how difficult it is to build a business, especially from scratch, I would have thought several times. That is, the probability of success is quite low. But entrepreneurship was always interesting: how to change some old processes, to make something more effective on at least a small period of time. In general, Problem Solving always inspires me - to find the problem and solve it. Even after any interview in Revolut, I was put by Advanced Problem Solving that I accepted as a compliment.
And in fact, it was not completely my initiative. I had pretty good partners. One already had two exits at the time of the founding of our joint startup.
- Initially, was it a Finmarket?
- Market Invois. It was a platform where it was possible to do factoring, pier-to-pier factor. But then it has already developed in the Multi-Product Fintech-Business.
- That is, they dragged you?
- They suggested.
- Have you persuaded a long time?
- No, not for long. But no one in the corporate parties - especially in such as Goldman - does not work.
- In the sense of it is difficult?
- Yes, it's hard. And this is quite such a rehearsal - the same thing is constantly. The experience is very good, I recommend. But to have a long career there ... In any case, I do not know anyone among my peers. Someone either goes to MBA, or begins to do something. Or just in another part.
-Probably, you can get there to some partner there.
- Can. But this is a very small percentage of cohort, which joins at the initial stage. This is less than 2%, as I understand it.
-Tell us how Kriya began [the British fintech company specializing in financing accounts, business loans and built-in financing]?
- It was obvious that Fintech was the next very large direction for innovation. At that time, 2009–2010, the trust of the banks was at a minimum after all Bailauts [the redemption of “toxic” assets. - approx. The Bell] that were made to save them against the background of the financial crisis. Banks had problems, for example, with a client service, because they grew to such a size that it was necessary to go through a hundred circles of hell to get a loan. And we had it all online, 24 hours.
- I wonder who these people who believed initially?
- First, they collected money for friends. I remember the sid-round. Now it sounds funny: we had an estimate of less than 1 million pounds. We raised 300 thousand pounds when assessing 900 thousand.
-That is, 30% gave to some friends?
- Yes, 30%. Now the CID round is 20 million.
- The world got rich. In any case, this part of the world was very rich. And how much is Kriya now and what is her fate? Is she Public?
- No, she is not Public. In 2021, when I left there, she made $ 25 million revenues and somewhere $ 5 million net profit. It cost about $ 200 million. We sold part of the Barclays. I went out in 2021 and after some time I started investing.
“Have you left the money from there?”
- Yes. Not megagans, but something happened to new projects. And she [kriya] still exists and is developing, my partner is still CEO. I am now much not involved in the operating part.
- Are you a sleeping shareholder?
- Yes, strongly sleeping. Because, of course, I have 100 plus a percent of my time to my current activity.
- Why did everyone go for you?
- It was a HOT category. Many good investors have appeared. The same thing is happening now with AI. Nobody was especially needed before Openai. Google bought DeepMind, the greatest company, for $ 600 million. What is $ 600 million for such an asset? It became so hot that everything related to AI raises money without any revenue and other unnecessary things, often according to $ 300-600 million.
“But is it insanity or is there anything behind it?” How to understand?
- Of course, most of this madness. We invest at the stage of series B, that is, it is already a company that has a product and already has about $ 5-10 million revenue. It is located on Inflection Point [turning point. - approx. The Bell] before scaling business and reaching later stages. Of these companies, where business is at least 3-5 years old, money is mainly earned by TOP 5% of companies. This is from 25 to 50 companies per year. That is, all the same, units, if by and large, watch.
Rounds of the series in are not based on fundamental data. It is done like this: there is a market that is formed or already formed, there is a strong team that builds good products, there is a product that is at the initial stage of scaling. And this can be proved by some single-digit [unambiguous] millions of revenue.
-Revenue is needed to say that there is a business model?
-Yes, [what] there is a business model, there are customers who are preferably more and more use the product. No one watches how the animated multiplier is coming. There is no profit there. Or even the multiplier from the revenue is the same. One of the interesting insights is that the assessment at the stage of the series in the innuman not very much predicts the subsequent profit that the venture fund received from this startup. Basically, today at this stage is from $ 100 million to $ 200 million.
- How is this formed at all?
-There is some kind of corridor of generally accepted assessments that the market itself finds. But in fact, this is just demand and demand.
- Revolut had a lot of component success, obviously. But those who use banks in Europe should understand that new banks, in principle, have huge chances to attract users if it is good to do.
- Even until today, when I periodically meet with potential investors, doubts are still visible. Then Revolut is too expensive, it is not known whether the profit will be there, it is not known whether they will give License.
- The main question is probably in License.
- Yes. In any even superuspre project, there are people who doubt the last in its success. And it is not known what should be proved so that they cease to doubt. We even check that our model choose a series in Revolut. Because, as I understand it, 90% of investors that Nikolai went to raise the series B were refused, because, firstly, the unit economy did not converge, at that time the company had a negative gross marjot. And the assessment was quite high - $ 300 million.
- When was it?
-This, in my opinion, is the end of 2017.
- Was the assessment then $ 300 million?
- Yes. That is, it was possible to earn 100x from the series V.
- And in 2021 there were investments of $ 33 billion?
- Yes.
- That is 100x?
- Yes. But this is a huge rarity. There were not many companies in the entire history in the entire history.
- Is this in general in the history of the venture market?
- A little, yes. More than 10x is 5%.
- Cool. We met with Nikolai [part -time] in 2018, when there was a billion. Since then it has grown 30 times.
Chapter 3. The capabilities of AI and the search for an investor at the stage of idea
- You said that this is not quite artificial intelligence. What is it?
- What is now called artificial intelligence is a certain approximation of intelligence. The model, which scanned a large number of words and from there found some interactions between words, some patterns. To digitize the words, you can use the so -called embedding. This means that we appropriate the figure, some kind of vector for every word. The most stupid example is to number all the letters in the English alphabet from 1 to 26 and build vectors.
“But it will not mean anything.”
- We need to do it: choose numbers so that words similar in meaning are relatively close to space. Say, the “king” and “lady” should be closer than the “king” and “apple”. Because you need to fix the meaning. They learned to do this now. And you need to understand the sequence of words. “Lisa eats pizza” and “Pizza eats Lisa” - two different opinions. Therefore, they learned to encode using the sinusoid.
The third is, of course, the context. There is a River Bank and Revolut Bank. The word Bank - “shore” and “bank” - means, based on the context, completely different things. And this was just the invention, such an architecture that, moving these vectors ...
- Is the vector a word?
- Yes, this is the word. She shifts them and searches for the so -called scalar work. For example, “fried rooster purr” - the words “fried” and “rooster” can often meet in one context, and “rooster” and “purr” are most likely never found. It is necessary to sew the vector so that, by changing the “rooster” and “fried”, the value turned out more than “Rooster” and “purr”.
-That 's that all kind people painted everything.
- If I remember correctly, in 2017, Google has an article attend is all you need. Based on this, an architecture was formed, which is called a transformer and has already become introductory for Chatgpt and other programs.
- Why does he not know how to predict?
- He cannot give answers to questions to which mankind still does not know the answer. For example: what is love, are there aliens, how to make a thermonuclear reaction at home? He studied all the texts, and this is a huge job for which a person will need 20 thousand years, and caught patterns. This model can now make a bunch of interesting things. For example, Summary-that is, read the big text and write 3 bulletinatomas, about which this text is about. It was most of my work in Goldman Sachs as an analytics. I read a bunch of texts and made a profile for the chiefs. Now it can be automated, which, it seems to me, is a big breakthrough for humanity. Used to automate physical labor, now you can also part of the mental
- The most promising areas if you start a business in this?
- There are a whole cohort of companies, AI Applications that take trained models and make the so -called tuning for a specific industry or for a certain context of tasks. Because the general model still sometimes says nonsense. She has the so -called hallucination. And if you, for example, try to automate contracts that should be Legally Binding [have mandatory legal force. - approx. The Bell], everything should be perfectly written. The share of improper text cannot be allowed there. Therefore, now there has been a bunch of companies that are trying to train the model on a narrower amount of data so that it already perfectly predicts the text in the context of contracts, for example. The same thing happens in medicine. How this industry is lined up, we do not know. It is possible that after some time Openai will invent a superinTelelex that you can train yourself for the spectrum of tasks that you are interested in.
- So you can use this?
- Yes. Suppose I work in a law firm and can download all my contracts, and the model itself will train, and I will have such a Copilot [Auto Fill tool].
- Startup is not needed.
- Yes. Then all the startups that are engaged in applications will not be needed. But when will this happen? One startup in which we invested is engaged in medical [texts]. I came to a physical examination, the system recorded everything that I said, and wrote the report automatically. He turned out to be rude, the doctor had to insert his amendments. It took more than 2 million medical examinations to train a model that can be more or less automated. The model is already very different from the incoming, from Openai or Mistral. If the superinTelelex is not invented soon, there is great potential for such artificial intelligence applications.
- Your bet: will they invent or not invent?
- Our bet, which will need for a long time for this.
- Long - is it 10 years?
- It is impossible to predict. This can happen tomorrow, after 5 years, after 10 years, may not happen at all. But we do not take this. Now from our cohort of the series, it is calculated which companies are the most promising. And often there are topics, as I told, about artificial intelligence applications. This is where we invest.
- There is AI as a big topic, and there is fintech as a big topic. How to come up with something that will become a successful business? It turns out that you need to take some kind of nascent business segment and think what he needs to work more efficiently. As a program that serves electrical supplies.
-Firstly, it is very rare when business from scratch becomes profitable. The percentage of those who win is very low. In general, from all.
In our [case], I repeat, a 5% chance of right 10 plus earn money. As for [that] how to find an idea - I do not know. It all starts with a problem. Can we solve it. There was such a topic in Crypto that there is especially no problem, but Krypto, as they say, Solution in Search of the Problem. That is, there is already Solution [solution], but it is not clear what specific problem he solves.
Transferring money at a normal course - this problem, as I understand it, Kolya had a personal one, because he easily understood where he was deceived, what an exchange rate is large/small and so on. This sunk into his soul, and he decided to solve this problem.
- You said that Peninonsky is underestimated as an entrepreneur.
- Yes, having worked with him for 18 months.
- Why is it underestimated?
- Probably he put it badly. I communicate with investors and sometimes feel: the fact that Revolut is one of the best companies in general in the history of European technological companies is not yet recognized. There is still no recognition of this level. All the same, many doubt, [ask,] when the profit will be, when will it be? I think that all questions will soon be answered.
- What is the criterion for the greatness of the technological company?
-From my point of view, this is still a scale, including profitability.
-Goldman Sachs will never die in you.
-Goldman Sachs was a very cool technological company in the 1970s, like all banks. They were the first to use supercomputers, the first to introduce Microsoft Office and so on, but over time, like everyone else, they were outdated. They have a huge number of Legacy systems [outdated methods, computing systems, technologies, etc.-approx. The Bell], on which you need to spend billions of dollars in order to change them.
Many people who wrote this code in 1970-1980 have already died. Plus, they had a lot of all sorts of murgers [mergers]: the Bank of Database with Clients has, but he died with some bank, whose scheme works in a completely different way. You need to integrate everything. That is, it is already a complex Legacy system.
- Thus, it is easier to build nearby.
- It's easier from scratch, yes.
- I can imagine how many requests are received by people who are directly related to investment from the series: “I have a project. Help find an investor. ” And I understand that, as a rule, this does not work. You can name a set of steps that facilitates people's access to money. In what cases is it better for a person to contact venture investors and when is it easier to take a loan?
-Firstly, you need to understand: this is Venture Backable Business or not. Most businesses are not created for venture. They solve a limited problem. There you can not build a company with an assessment of a billion plus. The whole game in Venture is to find companies that can grow to at least $ 300-400 million revenue.
At what stage is the money from investors - the Traction Trumps EVERYTHING [promotion above all things. - approx. The Bell], as we say. The more customers, the more progress is precisely in the product. Because it removes all doubts. The early stages are very much blurring you as a fauner. The less I can raise the early stages before there is already a product and there is revenue, financial results, the better.
“But it’s more profitable for you.”
- We need to watch it on the portfolio. A lot of startups does not shoot. More than 90–95%. As for the Founders, I can now raise a million pounds and give 30% of the company for nothing. Or I can wait for 2-3 years, when I already have customers, revenue, product and so on, and raise $ 20 million, for example.
- Often there is such a conversation: “I have an idea. I need an investor. "
-Yes, this is Pre-Eed.
-Still, at the stage of ideas you can?
- Of course, yes. There is an industry that makes a Pre-Eed round. This is usually Solo GPS ...
- Individuals with money, not funds?
- Or they have a fund. Solo GP - when the fund has one fauner.
- And the sowing means that there is nothing at all, even MVP?
- There is very little. On Pre-Eed-PowerPoint presentation and idea. Maybe even without PowerPoint.
- That is, at this stage you can?
- Everything is possible.
- Fine. Then what is needed for these people?
-We need a big idea how to do it differently, and you need a unique inside, why I know this. You can’t just come and say: I want to do what they do for money for free. This also often happened. And how to make money on this, we will see later.
“But the same thing works too.”
- Rarely.
- Social network Facebook.
- This is a little different. A completely different paradigm of communication. The track was cosmic. Everyone was sitting on him. I remember: I studied at Cambridge, and a friend studied at Harvard. He then called me on the phone and said: "Wow, we have such ..."
- In 2005?
- Yes. “We have such a topic turned out to be The Facebook.” She was also called The Facebook. Urgently her ...
- The?
- Yes, The Facebook. "We are all sitting on it for half a day, an awesome platform."
- Did she just come out?
- Yes. In Cambridge, she has not yet been, if you remember.
“She was inside Harvard.”
- I went to universities. First in Harvard, then connected Yale, Cambridge-in my opinion, another 6–9 months have passed. As soon as Cambridge connected, everyone was sitting on Facebook.
It is clear that in the end there will be monetization. Telegram is the same, very good indicators of Engage, which is measured from the monthly audience. In a good social network, 50% of monthly customers come every day.
-Seed and Pre-Eed-what kind of money is this? With an assessment or without evaluation?
- In my understanding, a good SID is $ 20 million an estimate now.
- This is a lot.
- Maybe smaller - from $ 10 million to $ 20 million, depending on the phantors. If these are repituers, of course, more, if it is a senior founder, then it can reach $ 100 million. We saw this.
- On Seed?
- Yes.
- There is still nothing, but already ...
- No, there is an idea, there is a Power Point Presentation, the outline of the product.
- It seems to me that this industry is doomed to be changed.
- Why?
- Because she is very closed. There is a lot of money and a lot of unpredictable intuitive solutions.
- Yes, I agree. Maybe we will contribute to a change in this industry. It may become more effective.