Why Many AI Startups Are Built to Fail (Expert Explains)

Añadido: 2026-05-15

[00:00:11]your hosts.

[00:00:23]Uh he's an academic researcher and international strategist institute of technology.

[00:00:49]Thank you so much for being uh our guest on Rise at this podcast. It's such an such a pleasure having you here, especially since you're here for a short amount of time. So, it's a great thing that we got to uh that we got to schedule this time with you for this interview. So, thank you for being here.

[00:01:07]>> Thank you so much. And to be all like short time on earth.

[00:01:12]Yeah. Nice. Nice. Nice seeing you.

[00:01:14]>> We only have today. Yeah. And our fate is in the hands of the chemical engineers I believe. So >> and the data scientists and yeah the researchers.

[00:01:25]>> Yeah. So our life is short but let's let's make most of it.

[00:01:29]>> Yeah. That's a good philosophy to have.

[00:01:32]So uh I covered a bit of your background on my opening but I want to know uh from your as from your side how did you get into this line of work? How did you become a strategist, an academic researcher? Uh how did that happen?

[00:01:49]>> Oh, that is simple. You grew up uh you do what you want to do most and that was for me like programming. So then like uh computer science was natural and then uh thank god communist communism ended in the Soviet Union and market economy started. So it was quite natural thing to do to go from the algorithms to logic of how people work, how like systems work etc. And that's what we did for almost like 20 years me and my partner building quite large structure doing that and everything else followed like even the AI work that I'm so much engaged today is coming from what I did in the university and so it's like most engineering stuff whole life like engineering how companies are working engineering how like huge like I believe my largest client for the strategy. They had like 350,000 employees. So, it's fun to engineer like that big things and everything else.

[00:03:04]Yeah. So, it probably was like kind of logical like in retrospect.

[00:03:10]>> Okay. So uh what I'm very curious about is uh what it was like growing up in uh in Russia during communism and after communism has fallen because I mean for a lot of us uh Russia is like a very secretive uh a very secretive and a very shadowy uh country. So for you uh being being raised there, what was it like especially as you grew up and became an adult? um you saw a shift in uh like you said in the the economy of knowledge just completely shifted. So if you can just talk about that a little bit. Uh, again that's what well first of all like when you're small everything that's around you seems to be normal and that's like how I feel like my kids are raised in Israel when we have the sirens and we go to shelter uh when rockets are flying but my for my like 11 yearear-old it's kind of normal that's only life uh no I don't I I'm not sure about it I haven't spoken to a lot about this but like that is something that is all his life is with that like in kindergarten they had this drills in the school they have the drills so kind of normal so when you live in like in the uh extremely totalitarian country today probably should go to North Korea to experience like that level of like mind control of every everything control of amount of lies that they take you like on history lessons for example about your country and you know that it's false uh that probably has no nothing that analogous nothing that analogous in in the in in in other places but like you grow in that so for you it's like kind of like your uh reflexes are to hide everything for example like I remember uh in Moscow subway reading Nabokov and preach inoculous well labok is whole Nabokov is inoculous it's not like calling for revolution or something. H but like but Naboka was not permitted not not allowed. So I had to hide the title of the book with the newspaper. No >> and and to see that no one see what I what I read and but well that was kind of like tricks. That's well when you cross the road you look to the right you look to the left. It's it's like that.

[00:05:39]>> Yeah. Uh and then when when like Soviet Union died obviously it was the best moment of our lives because uh you have free access to everything you can learn you can study everything that you want to study and obviously it was huge opportunity for like making money or doing doing business doing things that you want to do it's uh yeah I believe it's tremendous it's tremendous change and it was uh it was always like it made me happy very I I probably like when uh like in the Soviet Union when you grow up in Soviet Union in like special circles maybe groups I don't know but for like people around me it was pretty natural you are there until you get like to the third year of university. So you get like free Soviet education and then uh like with either with your family or even alone you go outside you have bachelor's degree already you get to the masters but you go outside and there was not much opportunity for that uh like but I'm Jewish so like you can do this there were there was not allowed for for normal people but there was a story about like uniting the the families that's the only thing >> okay >> but like it was was kind of like was programmed to to go like to other place to go outside but then like when Soviet Union collapsed then like wow capitalist you don't need to leave the country to go to the the free market and to normal life but you just have to wait and to build it here. So that's >> that's very interesting still but that's like history lessons. I'm not sure that >> everyone in in your audience will will benefit from that. Although there's a lot of things we can like talk hours and hours how the mentality how the how the mindset of the nation is determined by this paternalistic uh to authoritarian system and I see so much of that like in the in today uh Ethiopia and how people interact with me and some interactions that I have have even like on LinkedIn when the first comment that I get is usually about what will government think about something and that is like 2026 Ethiopia very different but that would be exactly what people will be thinking in like 1978 Soviet Union because government was everything uh so that's very interesting conversation but I'm not sure that that is where like your viewers >> no go a little bit >> it's just to understand uh you know because living in Ethiopia and living in Russia is completely different. So our experience here I mean we've gone through ups and downs in our history especially in a short amount of time there's there have been a lot of changes. So and Russia has had uh the biggest change of all but as you said uh we can talk about you know the main reason why you're here and it's talk about your work. So currently you're a researcher uh a researcher at the a lecturer and a researcher at Holland Institute of Technology uh in Israel and so your work focuses on uh AI and emerging tech technologies. So uh can you just uh explain to us what that means?

[00:09:07]I'd love that anyone will explain that uh to me even because we are tackling with my boss who is head of the school of computer science we are looking at many different things mostly like how AI impacts the like social contracts and like for uh Ilia professor Leven who is my boss the biggest interest is education he spent all his life and all his scientific uh achievements are in education ation of the people and how technologies influence education. So that is big part of what we are doing.

[00:09:43]Another component is I'm looking much because of my history and that I was like doing like business strategies uh almost all the life. So I'm thinking how this impact businesses >> and recently will recently started to to think a lot and I will probably publish on it how it impacts financial industries and banks and that is some uh review that some some data that I want to collect about how it happens here for example in Ethiopia that interesting focus uh but uh the like AI in society it's not that much about like how the large language models work all Do you have to know that and we are in a school of computer science so our students have to have uh at least uh better understanding of that that like people in general public because well they're engineers and that is the their focus.

[00:10:36]So like what I'm teaching uh in in the school is how to develop software using AI and that is not as simple as some people will think. there are a lot of uh we have like 40 hours and that's not enough to talk about this with students.

[00:10:54]So those are also interesting to me. Um and the angle that we might be uh it might be useful for us to explore is also like how everything like AI related influences the the startups and what what is going on in the heads of the startups uh when they think about AI and they try to like put AI into everything they do and uh that's like the phenomenon of like AI in Jera is something that I would be happy to talk about.

[00:11:32]>> Okay. So, yeah, since uh we are we are uh a podcast focused on uh startups uh and you told me that off camera that you did a presentation on uh uh on wait I had the title on the hard truths about AI and startups uh recently at Alex back in February. Uh so what what are the hard truths that uh most startups don't understand about AI? As you mentioned AI in general is that like uh an artificially intelligent ina like I'm trying to understand these things because we talk about AI like everybody uses AI everybody talks about AI AI has infiltrated every possible sector uh of the of the world. So what are the hard truths that you consider uh about AI?

[00:12:25]>> Yeah. Uh let me start with the preamble that uh every time I see like uh uh in a uh that's most most of the places in Ethiopia that I've have experienced with uh young people doing the u work as developers or like startups gathering together or like the uh podcast initiatives. I see a lot of young people with the bright eyes and with a lot of energy and I have huge respect for that and I believe that if I can help with with that if can I can help with like uh explaining a little bit what how how best to approach things what is the the best way how not to lose uh time and how not to do stupid things. uh that that is something that I feel that would be very good to do and I feel a little bit on a mission to do that. I hope it doesn't sound like too quirky but that like s quite sincere because uh yeah I I I really see like a lot of people who have energy and interest and just sad to see uh them not doing things in the best way and not optimal way or like losing focus and stuff. Uh so AI and Jera is simple.

[00:13:50]It's exactly what you said. that's trying to put AI on top of anything >> that moves and that is >> um sometimes it works sometimes it bring people millions of dollars in investments sometimes it helps people to land jobs uh if you go to the uh if you are just walking dogs but you say that my like I am AI walking dogs maybe it will improve your chances for getting job or for getting the uh dogs to walk.

[00:14:27]>> I don't know. uh there is it's a case a lot of investment uh a lot of investment companies will look to the next buzzword when they make decisions and a lot of customers a lot of like corporate customers might also be in a position when just the IT director for the bank just need to buy something AI so his boss will not ask him like why we haven't bought anything AI yet I heard of it I read about this in the newspaper like when I was flying a lot in my in my consulting practice. A lot of customers came to me, oh, our shareholder was flying on his personal plane and just reading newspaper. So he found like your article about strategy management and he asked me to call you.

[00:15:13]Now every newspaper is about AI. So boss is own owner shareholder is getting and like getting to the office and then talks to the IT director about like what do we do in AI? So maybe they will buy like AI and Jera but u doing that is um bad because not everything need AI in it >> because if it doesn't help the process >> absolutely >> become more efficient or help the customers uh get better access to information like customer service or yeah if it doesn't help the process then it's just it's just a flashy label like oh we are now AI powered It's absolutely uh and that's not well and I my experience is uh that any distraction from the the where you're should be focused on any distraction is bad. It's just like you you spending your resources and also in part it's like lying to your customer because you present this AI component as value and if it's not then kind kind of like not nice you don't want to be in this position uh but you are completely right uh in the very beginning because that is the the major thing actually like is the uh AI component uh helps you to to serve your mission if that is what you need to do for all the things that you need to do and also does it save money to your customer is your product better >> uh because of that and and and if it's cheaper because it's well we live in a real world and uh many startups uh um don't realize at least immediately that including any we call it cognific we like I would say that there's a book on that called co uh there is a term of cognification like making things smarter making processes smarter >> absolutely you can put like uh the large language model that is what powers like chat GPT and like other things the engine of those is called large language model if you put large language model into most of the business processes those will become smarter you may want like your uh I don't know like notepad on your phone to also think about like the topics that you want to discuss with it.

[00:17:44]So you will likeify your notes application. The thing is it cost money every like word every every word like every token cost money >> and and also like adds delays but like also cost money. So uh is economics there like did you thought of how much you will spend more and also like for many businesses it is new business uh new cost model and new uh new structure when like if your customer would love to talk to your notes application and will use it more and more then the price can easily overcome like for example you do subscription price so you have like fixed amount and with the cognification it will become like expandable to to so the the economics of the AI enabled businesses are different very different and uh if some small startup in in a will not be will not think of it it's no surprise even biggest companies in the world have that issue because like coming from the subscription to the use is a big deal and many customers are not ready for that. So for companies for even for largest companies it's still like a challenge but so economics of this uh projects of this uh companies the startup ideas uh can be sometimes very very questionable >> and u that is big thing to very big thing to think of very important I believe. uh other things will be like how unique will your preposition became because you added like AI did it help >> like uh for example I can have a lot I would have like a lot of conversation if somebody will if I'll be like an investment committee of the fund and somebody will come to me with the idea of like the cognified notes application for the uh phone I would really need to see what's the value there is it unique um many many things around it.

[00:19:53]>> I have a long list. Uh feel free to interrupt good and and come. Well, not not that long.

[00:19:59]>> I'm very engaged in this. So I just if I'm not saying much is because I'm like >> yeah I'm just happy to for us to be interactive.

[00:20:08]>> Um yeah and the other thing is soal like bulldozer effect. uh you may well some people may remember uh how Sam Alman head of the open AI uh two years ago was talking about the development of the uh CH GPA and the products that they do >> and he was quite emotional at the moment uh saying that if you talking to like startup founders or like people with with companies having having new ideas if you want to do some um something around our models when you are like compensating for their like less of uh lack of like some skills. So simple if GPT 4.1 >> is a little bit stupid and need to have some algorithmic addition or some so if your business is is assuming that a large language models will be at that level of smarts that they are today then uh what he said we will bulldoze you meaning that all money all energy all billions of dollars of open AI are went into making the models smarter and smarter and more capable and more capable. So if your business is on the same way, if you're like artificially making these models better, smarter by maybe adding some some software to that, >> then you will be bulldozed over when they will release like GPT6 because GPT6s will be so much smarter than everything that you do will be just history.

[00:21:55]And we uh there are a lot of examples of that. And the biggest one is if you will look to the publications about AI uh kind of like two year three years ago the big topic was how we get the data uh like if we have long document to be analyzed then and this document does not fit into the thing called context window. context window of the large language model is like how many words it can process like how big is capacity of of of the mind of the model. So it can process say 8,000 like when they uh launched it was 2,000 uh symbols.

[00:22:42]>> 2,000 symbols is not much. It's uh it's Yeah. So like >> 400 words.

[00:22:47]>> Yeah.

[00:22:48]>> Yeah. So at that moment the model was able to think over the sentence of 400 words and that sentence should also include your question to the model. So you are saying that this is the information that I give to you and this is the question that I have to you and then it will respond to you. So the qu the uh a lot of research a lot of work went into how we uh if the document is much bigger how we take one part of the document and put into the prompt then we take another part and put so there was like artificial things around the current capabilities of the models. Mhm.

[00:23:25]>> Today, when the model has not 2,000 tokens, but 1 million tokens, most of these technologies do not make any sense whatsoever.

[00:23:38]>> So, if all your business was built around the limitations of the current models, new models will make your business obsolete.

[00:23:47]>> That is the uh bulldo bulldozing thing.

[00:23:52]So the models will be smarter and smarter and smarter and most of the professionals in the AI world think that the models will be smarter and smarter and smarter. One of the things for example one of limitations that everybody was talking about year ago >> was models cannot work autonomously because models are bad in planning. Mhm.

[00:24:15]>> That was the models will never write good software because they cannot create good plan of of doing this work.

[00:24:24]>> Mhm.

[00:24:25]>> 12 months. Models today have excellent planning capabilities.

[00:24:30]>> Okay.

[00:24:30]>> They do it perfectly. They write big detailed plans. Then they read those plans again because like they have 1 million token windows. So they can have this plans build this big long plans.

[00:24:42]they can now use those plans and act on those. So just like one year and the capacity and the smarts of the models changed so much that any business that was built around like how bad it is.

[00:24:58]It's like very simple analogy of course is like human beings. If you're all uh if you only research uh humans uh who are like five months old, you will build big business around like helping this five months old humans survive in the world.

[00:25:16]>> Yeah.

[00:25:17]>> And you will be surprised to find that when these human beings get to like 15 years old or 15 year old, they have different capacities.

[00:25:26]>> They can navigate.

[00:25:27]>> Yeah. They they can do a lot of things.

[00:25:30]They can walk, they can run. So there are many more things speak.

[00:25:33]>> Yeah. Absolutely. And uh that is that is what the something a lot of startups a lot of companies died >> because they were like working for this like five months old models.

[00:25:46]>> So that is that's that's big thing. It's important thing.

[00:25:50]Uh another one that I was that I was uh uh thinking and talking a lot to uh to to that audience of the in this startup event in Alex was uh building the product is hard. It's just very hard job. You have to create the product that is supported. You have to understand your customer. You have to listen to your customer. You have to embed the feedback. You cannot just produce the thing and say that this is my product. There's so many things around it. You need to be able to to understand how to how how fast you change it. Do you have resources to change it very fast? Maybe like people were not buying it like from this very first second. Maybe like not thousand people will buy it but five people will buy it. And if you not update it they then your product will have very bad reputation of not update from like your earlier customers. You have to think about the life cycle of the product. You have to think about what product is. So like uh and most of the things will not be most of those issues will not be solved with AI. Chad GPT will not help you like to build your CHP can write you your uh processes but it cannot do it on your behalf.

[00:27:10]>> Exactly.

[00:27:11]>> So there are many things around like just building good products and those are very very hard things. uh that's why I would very hard to give like very generic advice but at times it makes sense to go to work for big company >> and understand a little bit how this company thinks about the products how they maintain the products how they develop the products >> before you start doing that well obviously there are >> it doesn't AI cannot replace the the human process >> uh >> the work that you do on the ground, the work that you do to understand your market, your customers, your product, uh what needs to be improved because that's all part of what you feed into AI.

[00:28:01]>> Uh first of all, the mo the biggest like thing will be if you want to be successful then you have to be unique.

[00:28:08]If you have like AI is not is not giving you like unique advice. No, it's not. it will give the same advice to the like >> another person. So if you have like your unique like insight, if you have your unique need, if you have like your your pain that you feel and like some other people of course pain because you you are not solving only your pain but if you have like unique view at those things it will be it it's it's very important but uh so AI cannot replace you in thinking of ideas but also AI cannot replace your discipline. AI cannot be the >> cannot work on your behalf.

[00:28:47]>> Uh it's you who have to understand that you are working 24 hours a day and you don't have any any life at all. and talking to uh the same in the same time when I was in in February I spoke to the students at the Bardar University and also in the Adis University in the the central one uh and I told them that guys the uh students who are in the software development and I told them that uh gentlemen ladies gentlemen uh the people who are in the position that they will soon be bachelors of this of the software development have uh two options uh and that's you're free to to decide one option is to study how AI can be working for you when you build the programs and the second option is die of hunger there is it's no no pressure at all no pressure at all you you may choose to die uh but nobody will pay a scent to the person junior programmer who is not using AI to the maximum because uh like Cloud code is good at writing like okay code. It's faster and it's cheaper than any developer alive.

[00:30:05]Uh no unless you want to get salary of like 900 $195 then it will be $5 cheaper than clo than clo but maybe that's not what you want.

[00:30:17]Um so uh the developer should be like running those things.

[00:30:24]>> Yeah. So uh my explanation to them was like yeah that's very very very simple alternative and I'd also told them that uh but in order to learn everything about how to command those like uh large language models to do work for you and to do to develop software for you because this is actually the I would say the only area where AI is already like arrived that is already working at working perfectly and you just have to master it >> but to master it will probably take 24 hours of the day. So what I told to the students is like year 2026 is year when you have no sex, no life, nothing. You sleep, you sleep very little.

[00:31:14]>> How to do the AI?

[00:31:15]>> In that case, in that case like in 2027 or maybe 28, you can start doing all that stuff. Uh but like this year learn study everything because the only chance to get to get uh in front of other people is just to put more and more time and that's what's kind of generic like any motivational speaker will tell you this but just feel it. It's it's really really something they cannot like lose that. So yeah, the same thing is like yeah, 2026 uh is a year without sex for that. That's like my big message uh to to people. I probably like put some posters on the street >> also. Yeah, I'll think about it. Yeah.

[00:32:06]And uh the same actually comes to the most like entrepreneurship stuff. If you want to do your startup until the moment when it will finally work.

[00:32:19]>> Yeah, that's bad news. Like no sex >> guys and no free time.

[00:32:25]>> Uh so that's and yeah I think that will cover like mo most of key things that I feel important in the context of startups and AI development. But the biggest is just avoid AI in Jera. Do not put AI on everything >> because people buy things without AI still. It's >> well a lot of businesses prosper without that.

[00:32:56]There are not that many areas where AI will do miracles >> will uh create like this um yeah the miracle sometimes like AI can bring this and that is good when you're selling things you just create something that was never ever impossible I just experienced that at some business meeting recently when I just uh came with the some idea that is easy to implement with the large language models and I see that for the people who I talked to it felt like wow wow that's that's that's fantastic but the number of such areas are not that big and uh you may have good business just like not with large language models there's an other AI as well like data science machine learning that's more like predictable but Yeah, that would be what I was I think that the most important to to talk about to people.

[00:34:02]>> Okay.

[00:34:03]>> And how was uh how were these uh these ideas received uh when you when you shared it with uh the AEX people, the students from Bahra University? I mean um did they seem to be aware of these things or was it completely new to them?

[00:34:22]because I kind of want to see compared to compared to the expertise that you have uh and what they know or what they have learned what's what's the gap uh in the middle.

[00:34:36]>> Very good question. I'm trying to see whether I know any technology that can get any feedback from the group of Ethiopian students and I believe I don't. Uh my father told me many years ago that there is saying in Africa speak Ethiopian and that is like when people are saying a lot of things around and never directly approach the the point and also like uh people are so people when people see somebody like that uh old and ugly telling them something they probably don't want to hurt that person uh for many reasons. I don't know why.

[00:35:22]So I hope it's humanism. I hope it's because love of the to the people. Uh but usually the feedback that I get is just like dead silence and I always begging like telling guys you think you are polite bec that you are not asking anything because you think you are polite but that hurts the most. So please please please it doesn't work usually. Yeah, I had that same experience uh at Adisaba University because I had to give a short lecture and then I was looking around the room.

[00:35:54]All the faces were like were they were just dead and I was like okay is it because you didn't understand? Is it because I was going too fast or is it clear to you? And it's just like still silent. I was like okay well if you don't have questions let's move on.

[00:36:11]Yeah, it's a bit tough to get people's uh to get people's ideas or uh perceptions of what you just shared. Um but yeah, at least I hope that it triggered something in their minds because >> uh on the other hand like the uh the main idea to which I just arrived in our uh podcast uh is very easily verifiable.

[00:36:36]If the call to like no sex in 26 will be heard, we'll have the deep in the new birth in the 27 and that will kind of be criteria. Did I like change the country?

[00:36:50]Did I reduce the number of new children >> dip then that will be that will be it.

[00:36:57]>> Yeah. Yeah.

[00:36:59]>> I >> but it's a bit intense. I mean like no fun, no sex, no no enjoyment until 2027.

[00:37:08]Yeah. I mean, good luck to the to the youth.

[00:37:11]>> Yeah.

[00:37:12]>> Well, there is some fun in having the islands and the good cars and I look at the streets of Adis and I >> rewards. These are long-term rewards.

[00:37:23]You have to sacrifice now to get these long-term rewards.

[00:37:26]>> Not that long.

[00:37:28]>> Yeah. I know people who did like who bought their island like in a year or so.

[00:37:32]>> Yeah.

[00:37:33]>> Everything everything is possible.

[00:37:35]>> Okay. How about how about your island?

[00:37:37]>> Huh?

[00:37:38]>> Your island? Did you buy it?

[00:37:40]>> Uh h No, I I'm not sure that I was ever in the mood for that. Although that's a good idea.

[00:37:51]>> Yeah, >> that's a good idea.

[00:37:54]Yeah. uh thinking about how bad air is in Addis having island and no one ever will burn any incense and no one ever will burn any fire and there will be no cars.

[00:38:09]>> Uhhuh.

[00:38:11]>> That's >> just bicycles and boats.

[00:38:13]>> Actually, islands with no cars give bring interesting effect. Uh you may know like uh near Istanbul uh there is an island called Buocada >> and they have no cars there >> and you can imagine the smell.

[00:38:29]>> It's >> because they still they still they still need to travel.

[00:38:32]>> Yeah. Horses and donkeys. Okay. Yeah.

[00:38:35]>> That Well, that's good fertilizer.

[00:38:38]>> Uh >> looking for positive side.

[00:38:44]>> Yeah. Good fertil. Even if this air smells bad, it's good fertilizer.

[00:38:47]>> Absolutely.

[00:38:48]>> Yeah.

[00:38:50]Okay. So, um All right. We talked a lot about the the hard truths about AI and startups.

[00:39:03]And speaking of startups, uh earlier we mentioned uh the startup proclamation.

[00:39:10]So since you're a researcher uh you have that uh academic insights into these things. So what was your idea about uh what was your first impression on uh the startup proclamation?

[00:39:27]Um I'm not only the well current I most of things I do is research but for a long time I spend my days writing like policy documents for the government of of Russia and we did a lot in the field of the innovation development and also we had we spent a lot of money of the taxpayers uh traveling around the world and getting the uh best knowledge about what's going on and what what's working and what's not working. And my uh feeling after all of those things was that uh well private entrepreneurship cannot be replaced with anything and and the signals from the private markets >> is also nonreplaceable by but any by anything. So looking for the uh efforts that are done in many countries of Africa uh in terms of like development of the innovation ecosystem and things around that uh gives me feeling that people that for many people it's easier to just try to do some government initiatives >> and that would not that usually that creates the um very wrong motivation for people.

[00:40:51]People are playing to it's it's normal.

[00:40:54]We all play to the performance measures that are given to us. It's that's a very well-known story that in like in one American city uh the firefighters were getting paid by the number of the fires that they put down >> and number of fires went up and it was proven that people were like starting some fires. some five some like firefighters that were like like not working today were starting fires because it was helping their colleagues to get better uh indic better better numbers. Yeah. So if you create set of incentives that is artificial and that will be in every like government supported program because the uh major incentive is just money from the investors and that is that is how you you measure or just improve the quality of life to the people. But as soon as you start to like to set things of how many startups, how many startup competitions, how many projects started by the graduates of this university of that university, people start playing to that and you will see just the pretense.

[00:42:09]>> Yeah. And you will see just pretense. So you burn money for all that stuff. So a lot of things in the like government uh like focused and government first uh like initiatives are like this and it's easy to think that for example like well Russia took a lot from Israel on how Israel built its ecosystem and but it was like misinterpretation of so many things. Uh I know some African countries are thinking about France as the example for this there. I I would have two issues. First of all, it's impossible to underestimate the level of the free market development and the richness of everything every player of in the free market that exists in France. It's hu big developed economy has everything >> and also I will just ask like what are results of the like France uh innovation system I am happy with the many things that the country produces but is it a pioneer in the innovation economy >> I would have some doubts around that uh and also like those initiatives the government government funded initiatives is like things of all very old past. We can all say that like even the biggest innovation system in the world and like some officials love to say that but even the biggest innovation system in the world that is American of course Silicon Valley was created by the government because many people will say that this is created because the ministry of defense was giving the contracts to the companies created by Stanford and Berkeley uh graduates and but that was like 50s it was very different era And things were very different. And still you are talking about the huge market economy where everything is already developed when the uh most of the infrastructure is built when the most of the ecosystem is just the ecosystem of the normal market economy. So you only build some parts because other parts are already supplied. But in most of most of the government programs that are known to me will have uh a lot of weak sides. Mhm.

[00:44:35]>> And uh so I think that startups in in in in Ethiopia could be should be less focused at least shouldn't be focused using that uh like incentives and everything maybe but not like uh not build your strategy or around like being in these startup competitions. I've been uh in the I've been judging the startup competitions for longer than I would love to admit, but uh it always almost always a joke.

[00:45:14]It almost always something very very strange and nothing to do with the like investment opportunities for the startups. Nothing that proves that the startup ideas is good. Yeah, that's true because uh that there's that uh gap in investment but uh I think I don't see it as all bad because there's a side of it where I mean I I have had personal friends who have uh started their own uh their own startups. I've had friends of friends who have done the but but when we have these chats about what it's like to have a startup in Ethiopia uh especially when they're operating here the issue is that uh fiscally they don't have uh a structure that fits into the current uh tax model because for them to de from the time that they started uh their their company to when they reached the market and they get their first customer etc etc. There's so much time that that has passed in uh in development and in uh customer outreach that they can't really get to that stage of like okay now we can pay we have an income so we can pay taxes. So I think first it's was bit was done to find a unique uh a unique way that they can fit into the tax system without being penalized because a lot of companies I mean normally companies have to have to pay taxes I think by the after the first year they have to pay taxes so like for them when they have like when it's a business you can do that but when it's a startup and you have R&D and you have uh when when you're experimenting in a very new uh market. It's >> if I will put like the strategy hat, I will tell you that uh that is excellent example. Most like almost everything that you're saying are like moving us and pushing us into very very proper direction. something that I probably missed when I was talking about this uh like the uh well I hinted at this but not did not like put it that concisely.

[00:47:32]uh like uh when we uh when countries develop like innovation things and want to develop innovation economy in the functioning free market system they only have few models uh of what is usually in the government like mandated uh innovation reform. Your example is not actually not about startups. Every company that is starting its business should not be under the condition that you are either you make money in the first year or you are stealing money from the government so we will do audit over you. So what I'm saying and actually like this idea of selecting some group and putting this group into the different like tax regime >> it is something and actually like same story about like special economic zones.

[00:48:25]It all comes from the communist China and that is about like we don't make life okay for everybody.

[00:48:33]>> We just select few things because actually that is what should be reformed in the tax system for the whole country.

[00:48:40]>> Any company without any designation should be in the position that they have as long as they want for like investing and things and like the tax system should be changed. Like another thing that they had either in startup proclamation or in other document before that the company could have no like legal address. There was a requirement that you have legal address that you paid whether you have office physical office but somehow decided yeah well those companies might have just office but like not physical presence or like work in the some co-working space. So we allow registration of the company that will have just PO box as the address.

[00:49:23]>> But why that is not available to all companies >> and uh so and all other stuff actually like this idea of the business license when you register the company and you and they need to find you in a list of 1,500 or something what is your business and match you with this.

[00:49:43]>> This is not category.

[00:49:44]>> Absolutely. It's it's called like business license.

[00:49:47]>> Yeah, >> that does not exist in many most places in the world. That is not how market economy is working. You don't have ministry that is knowing all kind of businesses. It's impossible. It does not work this way. Businesses have like mixed >> ideas about thing and you it's counterproductive. Yeah.

[00:50:07]>> So when they will be like releasing this for the like startups, that's not the solution. The solution is just to do a little bit more changes.

[00:50:20]I can >> add more flexibility. Yeah.

[00:50:22]>> Yeah. To general rules like just make business easier in a country. It might be politically harder to put it like through the parliament for example because you can have like we're doing this for startups. You people probably don't know. It's like very good thing.

[00:50:36]So for them we will do this and you and people in government might think that we will do it here then we'll show to the parliament that it's okay like the world didn't end it. Uh maybe I I just don't know the the whole like calculation and on the other hand I think that it's hard to think of like economic development as the task that should be done like in 50 years with the small steps. people will die because of the lack of opportunities about the because of things were not provided. So I I I really I'm not sure like any country innovation policies uh mandated by the government is uh to me problematic. I probably just know too much in that field. I don't see a lot of the successful projects of that just like make markets work >> and when the markets and one of the things that probably you heard a lot about lack of investment and lack of investment funds >> but uh important thing is to like call by name a lot of people understand this of course I'm not the smartest not by any measure uh but the uh number of exits are also very limited exit is when the uh When the fund when investor buys the company, investor usually should know how investor will sell this company >> because that's how like investors that's how the investors buy their islands.

[00:52:06]>> Mhm.

[00:52:07]>> They buy companies, they sell companies >> and if there is no market to sell the company, nobody will buy the company.

[00:52:15]And there could be probably very little market for the buying the companies when the what 70% of economy is state-owned >> enterprises state-owned enterprises have very little incentive to buy the the companies after the like fund invested and developed it. So why startups don't have that much investors? Because investors don't have that much like options to sell those companies later and you should think about it always and that will like answer a lot of questions. So to build like a lot of investment funds in Ethiopia, you not only need to like issue the proclamation on investment funds, you just need to create the the the situation where it's possible and without it just not waste time and energy and things.

[00:53:11]So that is like more about like the development of the market economy as a whole will raise also like new companies like also startups is well it's a good term but us like any company that just created is a startup >> is a startup yeah >> so and the uh the thing that I will be also saying is that and I don't know I I I don't think it's something that can be solved and something that can be fixed or maybe even something that should be fixed. But in the uh a lot of people who are who are thinking of the startups in Ethiopia, they have their like context their like frame of reference is western world.

[00:53:56]>> Mhm.

[00:53:57]>> So they are thinking about some innovations for the that will that will be great I don't know New York but that is not the biggest challenge in Bayard.

[00:54:07]>> Exactly. So you may be thinking about much less fancy things of course like AI mobile first all lovely and wonderful >> but if I would be in the position to like direct startups or new companies >> I will have like thousand companies building the service over the USSD. USSD is the message exchange with the chip phones.

[00:54:38]>> Yeah.

[00:54:39]>> Because like 80% of the phones in the country >> are those chip phones.

[00:54:45]>> Exactly.

[00:54:45]>> And you can do a lot. You can probably do like chat GPT over USSD.

[00:54:51]that will be something that where I will be looking >> for and like the medical assistant over the USSD and I'm sure that banks are aware banks are thinking because well for them real money >> let's do USSD >> yeah but for like fancy people who graduated have bachelors from the University of Texas and coming back to the country and that's commendable I have enormous respect for young people who come back to the country uh and that is That's wonderful. But the terms but your like uh your frame should shift. Go to bard, go to like smaller villages, talk to people, >> listen to their pains and and look at the resources that are available to them. Still some people will not pay for internet because it's expensive >> and they don't even have coverage even.

[00:55:46]>> Uh some people will not have coverage.

[00:55:48]Most people will have coverage actually most of people have coverage. Uh >> probably at Telecom should be more transparent about like what is their coverage but you you can get some information like this. Uh anyway, just like look less for fancy things >> but more for things that will solve pain >> and uh yeah that's that's probably the the biggest they can be >> and it's a and it's a huge market because I think outside of Adoba it's hundreds of millions of people that uh that need that that need these services but just because of uh localization for example, language or uh technology access issues, they're just they're just forgotten from uh from the mass demographic.

[00:56:40]Yeah, I believe that the um you used very um well you haven't like finished the full sentence but what I will hear is like uh and the way how I will frame it will be uh not like people are forgotten but by something no startupers startup people they're forgetting about the key audiences it's very I think It's important to put like the spotlight on people who should be thinking about it.

[00:57:16]Uh maybe other people also and also thinking about the financial models like business models people poor they don't have enough maybe still they have like tenure for this. Maybe you can prove value and you can easily explain value and you can like build a communication properly that will explain value to people. many many many things are different in in a country and there are a lot of country specifics and you have to think about it when I hear for example uh it's very different it's not it's not that much less high-tech although it's still technology I'm thinking about I hear a lot about bajases that should be on electric drive I I I love that idea uh but the most people do the simplest way simpler or simplest way they just put the uh battery and electric motor into Bajage.

[00:58:13]>> Mhm.

[00:58:14]>> What I am thinking always that if you will have like swappable batteries like I have in the electric bikes >> it will be different. It will be much better solution because imagine like the village again near Bodar. My family is from Bardar. So uh for me it's like I've been there often and I probably understand this. So look at the think about the village in Bodar or even like about outskirts of Bodar where most of Baj drivers probably live. Do they have good electricity?

[00:58:43]>> Can they recharge the big battery >> or it make more sense to have like the uh two like charging stations with like solar or something in the very center where at least there is some electricity and very regularly. So you can can recharge like small battery and usually like budget also something that uh well but I I spent like 20 years in min strategy consulting so for me these things are natural I spoke to drivers of by judges I I understand the distance that they travel it's not big actually so you don't need that big battery >> so if you have like place in the center that have rechargeable batteries it will be okay for for most of the things that Pajis is doing but and also you but you will have it connected to the electricity. Yeah.

[00:59:33]>> And there are many many things like that. So like look at the local uh local situation and think about what's working in the local uh situation for the people who live there and uh so this is this the the biggest thing that I believe might be sometimes lacking and when people when I I don't know I I I receive on LinkedIn I see like the gatherings of the startup people in in Ethiopia in Adis And I don't see that much conversation about low technology about like no technology what we can do for like I don't know for avocados >> to be processed locally what we can do in this or that what we can do to help uh old ladies with the eye problems mean there are a lot of opportunities and there is good chance that there is much more money and much more business and much more energy in the things that uh much more low tech.

[01:00:36]So interesting.

[01:00:38]Okay, that's very interesting because uh there was this uh uh UNDP program that uh sponsored innovators from all over the world and they one of them was uh uh like you said earlier was they were converting the bajudajes the uh fuel uh the fuel budajes into electric bajudajes and that was like a proper full-blown service where they took them in changed swap them out and they were doing the customer the the customer service the follow through uh doing the maintenance and everything with them. So that was interesting to see. And then um there was this other company doing um in Latin America I forgot which country but they were collecting uh used batteries putting them together and to make uh uh u like big uh battery cells. So that could be used for the home in case like there's a power outage. They could use that for the home. They could use that to replace the fuel cell in the electric car. So it was just it was very fascinating to see that. So it's I actually do agree with you that local entrepreneurs have to step out of Adisawa because Adava is just a one-of-a-kind.

[01:01:59]um it's kind of an oasis in a country that's missing a lot of you know a lot of solutions for its people. So I do agree that they have to step out of adis they have to look into uh low tech solutions and they have to just be closer to to where the people are. I think that's that's the main thing that I would take away from this.

[01:02:24]All right. So um I think um in conclusion I mean for one of our last uh questions it's been it was a very fascinating uh discussion by the way like I think I spent most of the time not talking because I was just very attentive to what you were saying.

[01:02:48]>> Ethiopian of you.

[01:02:50]>> Uh no no I'm I'm very non Ethiopian. I mean like if I want to intervene I will intervene but I was very uh I was very fascinated by what what you were saying.

[01:03:01]So what would you say is um I'm kind of leveraging your experience here. So what would you say should be our next step in um in growing the startup ecosystem in Ethiopia?

[01:03:18]uh if uh if there is any opportunity and I hope it is I'm not that much close to what's going on politically and sometimes big reforms they need to be executed at the moment when the politicians are safe and they feel themselves safe but if that moment will come as soon as possible days weeks >> huge economic reform uh not for startups specifically but just like do everything just remove all the barriers that exist and very obvious one is like this business licensing just destroy that like reporting u we don't have like in Israel there is no business license >> really how does it work >> you just you just you start your company no you there are two things registration of the new company.

[01:04:18]>> Uhhuh.

[01:04:19]>> And license and in Ethiopia it's two different steps.

[01:04:22]>> Uhhuh.

[01:04:22]>> You register the company >> and then you >> only when you or or right after you have to choose the license and that will be the only activity that is permitted to you.

[01:04:31]>> Mhm. Uh in many countries the civil court says that enterprise can do whatever it wants if not creating like employing chemical engineers to creating bombs like every legal thing that is permitted by so registration means automatically getting business licenses for everything >> and why it should be like in some other way because in the like because I I I didn't like think that we will talk on that specific. Maybe something in me feels uh this way. Uh because also when you have like this business licensing, it means that you need to have like a million bureaucrats who will be watching if the company is doing exactly what is in this business license because what if uh you don't have business license to sell shoes and you started to sell shoes then I'm shut down.

[01:05:25]>> We probably should kill you.

[01:05:26]>> I'm shut down.

[01:05:27]>> Yeah, absolutely. We should do something with you.

[01:05:30]>> Yeah. So just remove that that is does not exist in France. It should not exist in Ethiopia. There is no reason uh that and reporting and all these requirements like if you register the company in France if it's small business you probably report electronically once a year on your taxes and that's it.

[01:05:54]>> You don't do like I believe here it's like either monthly or bimonthly reporting. Yeah, it's >> and you also need to have like the accountant because your report should be signed by a person who got the education. If you have small business and most businesses small look at what we see around like the uh the size of the businesses well even the bank many banks here have like less uh sales than uh McDonald's in the small American town and this McDonald's like files taxes like once a year >> and banks are probably doing this every day because that's how banking system work. uh I'm talking like simplify not no not I'm not talking about like remove the banking banking regulation I have own reason and they they should be very tight >> but for general businesses it's not about startups what I'm saying I'm trying to say that make everyone that is like be business in the beginning make it very easy for them for businesses that are bigger like well think about this >> for more requirements that would be normal >> if like as you were saying for small businesses make it easy to start and to get into uh making money, making revenue, paying taxes. for bigger companies maybe add more requirements like maybe uh ask okay what is your domain of uh uh what what uh what sector are you are you practicing in or what uh I mean >> I would say mostly it's it's usually it's it's more about accounting requirements because uh limitations with the sectors I I I I really don't know any like successful economy that has any anything there's no no restriction and uh for some reason even like in the in Russia after the uh collapse of the Soviet Union we never had this uh thing with the business licensing. So uh because like we were Russia initially like when the market economy in Russia started Russia was so closely following every step that IFC wrote it was very very very disciplined uh implementation of the free market reform something that like a lot of intellectuals here are saying that like evil but don't think it's true. Uh so but >> why are they saying it's evil?

[01:08:17]Oh, that's very different conversation about the values about the what our country stands for that we did this for 3,000 years so we shall do it like for next 10,000 years until the we will all die etc. It's uh it's >> extreme hardcore discipline. Okay.

[01:08:36]>> But that is something that I believe like 80 oh 60% at least well judging by my LinkedIn like 70 80% well uh follow yeah so do not look at startups as some different animals. M >> all small animals are the same.

[01:08:56]>> Mhm. Whether it's like high-tech or if it's like just person growing the I don't know avocado on the uh um one hector small or like 10 hectares but very small business look at this treat them the same way and then startups will also prosper because those companies will grow up free market in free market economy companies will grow up and companies that are on a competitive market they are buying startups because you have to be on the competitive market. If you're a state-owned company, you have zero incentive.

[01:09:32]>> Mhm.

[01:09:33]>> If there is any idea that like Ethiopian Airline, if Ethiopian airline think that something is a good idea, >> they'd rather build like department of this good idea and hire people to it than buy company from the market. Why they need this? It's more expensive probably because usually companies are buying startups because they want to save time.

[01:09:57]>> Yeah.

[01:09:57]>> They pay a little bit more or much more but they save time because they got the readym made team.

[01:10:03]>> Yeah.

[01:10:03]>> And large government organization hate this because they need people very disciplined. So they need them to grow up in their system and they need people who know that the whole life of these people depends on the decisions of and that's not how startups work. So only development of the free economy will improve number of the buyers of the technology startups only more buyers more investors are willing to buy it at the moment and sell later. So all this like works together. There is certainly that's a right word ecosystem. Uh and I will be probably the last person to say that's a bad word but it doesn't look exactly how governments tend to look at this. Mhm.

[01:10:47]>> In order for not just startups but for companies to thrive, the free market has to h has to has to be there. Has to be has to have the players and also the the companies and the investors uh in it so that it can be a fully functional system.

[01:11:04]>> Yeah, it's very hard to imitate free market. That's what China is trying to do all the time. like having like something that will imitate free market but not be free market because free market uh brings free thinkers, free market brings like free political forces and that's Chinese Communist Party cannot afford it obviously. So they have this illusion and this imitation and uh we see result of that every day like number of the electric car companies that you see on the street >> is a result exactly of this game and that game is played uh if I will like little bit more details that may help a little bit like why that so many brands of the Chinese cars and what you have here is even like less than I don't know 10%. You go on YouTube, you can see the Chinese exhibitions of the cars and how many brands are represented there. And that's like tens and tens and tens of the brands. And the reason is very simple.

[01:12:06]>> There is the competition between the regional committees of this of the communist party of China >> who will build the best electric car company. So every city, every city communist uh city committee of the communist party will try to have their own and as they have power over local banks, banks all banks in China government owned and people in the like party bosses can full control over that or maybe partial I don't know and also they have control over what is purchased by the uh city administration for example BYD great company. Mhm.

[01:12:47]>> Uh I believe 100,000 electric buses was what was ordered by the city I believe it's city of Shanghai that started that company.

[01:12:57]>> Mhm.

[01:12:57]>> So on on one hand they initiate a huge buy of the products.

[01:13:04]>> Yeah.

[01:13:04]>> And then like BYD was also bought by local police and then it was bought by the army because they had some connections etc etc. And then you push your local banks because they're also governmentowned to finance this company.

[01:13:18]And then you got like huge scale and scale brings down cost.

[01:13:23]>> And while BYD is like very respectful company, they are in the energy storage.

[01:13:28]They are very much very advanced. Uh but does the world need 100 brands of the Chinese car? Does China need 100 brands of Chinese car? No. We have that experiment. That experiment worked interestingly in solar in solar panels.

[01:13:46]Again, there are like hundred of the brands of the solar panels. That was previously uh the love of the communist party of China. Every city should have their own like manufacturer of the solar panels. all everything that I'm I'm telling you like this stories because one probably it's interesting to see like why there so many but another thing is like you don't imitate the market >> and like what happened that with the solar panels solar panels now are extremely cheap the profit that the Chinese companies are making like from the components to the panel is considered to be like five to 7%.

[01:14:22]>> Mhm. And that is actually like why I am very sad every time when I hear about the idea of the local assembly of the solar panels. If you bring all the components here and develop the solar panels, there is no chance in the world that you will fit all your like investment into the 5% profit. You probably will have like 50% profit. So you certainly will make solar panels much more expensive than Chinese. It's just it's just math. It's just math. uh but uh what I'm the whole story is about the like you do not imitate free market forces uh they also did it even before before the solar panels they did it with real estate every city was ordered to build a lot of real estate because population is growing in terms of >> what we have today uh and for 10 years in China we have huge crisis with this real estate they built m so much more that anyone needs and that is like they imitate groceries >> abs that's something that we all see because uh like even Chinese communist party cannot like hide this that is that means like you do not substitute free market well unless well communist party Chinese communist party have the reason they want to exist they don't want to be killed all of them they want to be in the power so for them it's like existence fight for existence but if you don't see yourself like fighting for existence and or if you think that your existence is free market economy then you don't imitate those things and building like very special section of the economy that will work at the normal uh rules but others not it's all limitation I would not I would say like not play imitation game but that is like on the level of the policy on the level of startups just be realistic look around work with the customers and never make like incentives that can be created like today morning just for the like government policies or something never make those incentives rule what you are doing because long term incentives are not that important >> today maybe it's good you will buy new car because you got some grant or something but long term no it will substitute your uh market uh intelligence you will not do what people need you will do what the uh people in some committee will need and that is not long term so everything about the like um developing innovation in in a country it's usually done naturally and substitution of any like government initiatives to free market is not working that that's what I will say >> okay I think that's a very very excellent way to wrap up uh this interview unfortunately this is all the time that we have so Mikuel thank you so much for giving me this time. Thank you so much for being here. I hope you enjoyed your stay in Adis. I think you have a few more days left before uh your departure.

[01:17:28]>> I do.

[01:17:29]>> Uh do you have any plans uh to >> Oh yeah, I have very shortterm plans.

[01:17:33]>> Yeah, >> very short >> to enjoy Adis cuz there's a lot of >> the place I want to enjoy is couple minute walk from here. That's >> biggest plan.

[01:17:41]>> Okay, fantastic. Fantastic. I hope you have a great time. Uh and once again, thank you so much for being here. Thank you for inviting me and hope it was >> Yeah.

[01:17:51]>> helpful.

[01:17:51]>> It was very insightful. It was very insightful.

[01:17:54]>> Yeah. Thank you. All right.

[01:18:01]It's a wrap. Bye.