# "Most dApps Betray Privacy"- Lefteris Karapetsas, Rotki. w Peter Farbey | Web3Privacy Now Academy

- Speakers: [Lefteris Karapetsas](https://streameth.org/speakers/lefteris-karapetsas)
- Channel: [Ethereum Cypherpunk Congress](https://streameth.org/ethereum-cypherpunk-congress)
- Date: 2025-10-09
- Duration: 22:49
- Topics: W3PN, Web3, Privacy, Now, Blockchain, internet, Interview, Peter Farbey, Academy, digital right, digital era, cypherpunk, hackathon, hackers, hackers ethics, web 3, Story, crypto, data, protection, ai, research, developer, security, lefteris, Karapetsa, rotki, Education
- Watch: https://streameth.org/watch/yt-maF8fcdDqfs
- YouTube: https://www.youtube.com/watch?v=maF8fcdDqfs

## Description

In this Privacy Academy Interview, Lefteris Karapetsas (Founder of Rotki) dives into the challenges of building privacy-respecting software in Web3, the tension between transparency and privacy on Ethereum, and why financial self-sovereignty matters.

This interview is a must-watch for anyone building in Web3, caring about financial privacy, or questioning the current state of dApps, AI, and decentralized infrastructure.

Lefteris Karapetsas: https://x.com/lefterisjp
Rotki: https://rotki.com/

00:00 – Introduction: Who is Lefteris Karapetsas & what is Rotki
01:25 – Why he built Rotki: FOSS, financial transparency & privacy
02:11 – What is Rotki? How it preserves data
04:00 – Dapps building challenges
06:01 – Dapps building approach advice
06:55 – On decentralisation and privacy
07:46 – How to survive and be sustainable
09:25 – Why more projects should charge users
10:35 – Dilemma of privacy & transparency in Ethereum
13:42 – Practical advice on how to be private online
15:15 – How does the internet of the future look?
18:23 – Hot take on AI. How to use it privately?
21.49 – Final reco: piece of culture to enrich your mind

...

This conversation is part of Privacy Academy Interviews – a peer-to-peer knowledge project exploring the most important ideas in Web3, privacy, and digital freedom.
Learn more: https://academy.web3privacy.info/p/interviews

A W3PN Media House production.
Directed by Federico Marchi https://x.com/BabyBitProd
Interview by Peter Farbey https://x.com/0xfarbey

Let privacy be with you!

## Transcript

Uh so my name is Leis Carapetas. I am the founder of ROKI portfolio tracking and accounting tool that respects people's privacy. [Music] >> Hi. Hello. Happy to see you. So uh we have a super express interview because you have just 15 minutes. I'm sorry for like catching you that late but I really wanted to brief you to discuss you discuss with you like lots of things that um uh you are familiar with. So but like briefly briefly let's start like for the people who don't know you like a bit of your background like one minute how you ended up as a rock founder >> like I have been uh in Ethereum since uh 2014 um working for the Ethereum foundation here in Berlin uh before the launch of Ethereum uh was working on the C++ client and the solidity compiler and then later after the launch we continued is working on on the ecosystem until I switched to um standard slit that made the DAO and I'm sorry everybody saw that uh and um then I was a technical uh lead of the rider network which is like um the lightning network is for bitcoin that was supposed to be for ethereum and during all that I wanted to account for whatever it is that I'm doing in in uh the crypto case and there was only bitcoin.tax which is an centralized website where you just give all of your financial data and they will tell you how much you owe in taxes and that seemed insane to me. So I started at two uh command line tool at first that's uh became rokillator basically as a solution to the problem of how can I do accounting without giving all of my data to some kind of centralized sense. Okay, beautiful. I would like to hear more on rod key but we just have just 10 minutes. So maybe super briefly you can say about rod key and then we will jump to like general decentralization questions. >> Yeah. Uh I mean uh rod key is a portfolio trading and accounting tool uh that keeps everything local. So it's a good old local application that you just download um and use. It will connect and pull all of your uh transactions, decode them, understand what it is that they are doing. It allows you to interact with the chain. So, it's like a kind of a manager of everything that you do um in crypto all from inside your local system. The main selling point compared to every other uh alternative is that it's uh only on your local system. If something happens to me, you still have it. It's all open source. someone we can just continue it. Uh I cannot stop you from using RO key and I cannot take your data from you. It's all local. I cannot say okay from one day to the next just the company goes bankrupt and then doesn't exist anymore. >> I mean you can still use a lot because it's just an open source tool that you download. >> Yeah. Yes. It's a really wonderful tool and uh maybe next interview we will discuss more about the challenges you had to overcome building that because it's very interesting the builder path. Uh but um I wanted to talk with you about like a bit more about like decentralized apps uh and like the problems of decentralized apps in general now like why not so many when many apps why why like what's your perspective on that and maybe you can give also some advices to builders trying to get build some deps like how um they can do that. >> Yeah. So why don't we have many dancer? Why don't we have many reals? >> Both and it's good that you separate that. >> Yeah. >> So let's discuss in this and that side. >> Yeah. I actually think that we have enough apps like both dabs and centralized apps. Uh there is there's quite a lot I think. Um but there is very very few that are actually dabs. Well, just everybody calls a DAB nowadays anything that has to do with crypto which has killed the name because DAB is supposed to mean decentralized application. And in the end, what dubs most of them are at the moment is just like some centralized server running a front end running a code that just interacts with uh a lot of indexers and maybe at some point with the chain. But that's not really a because the original vision of decentralized application that we had when we started Ethereum was this you know the mist browser. It was like this local application that runs on your machine contacts the chain locally and shows you the front end for each application. Let's say it didn't exist back then but a unis swap. It's just all interacted with from inside um your your local system. Now why don't we have that many doubts is a good question. It's purely that it's actually very difficult to do it. It's it's so the builders path is very very difficult. Um so advis on builders on how to do that. >> Yeah. I mean you had this path right and you had certain challenges. Maybe you can share just challenges that you had on the builder's path so that other builders could avoid it. >> Yeah. Uh I the main challenge is don't do it very difficult. >> That's a brilliant answer. Don't do that. just forget. >> I mean like it's it's a challenge. It's not going to be an easy way. So like if you want to go for for for just easy money, just don't build real like makes no sense. Um but if you are here for actually building solutions, then um um go step by step and don't try to do everything at once. like don't try to solve all the world's problem in this one application that you're building. I I've seen many builders fail because just maybe getting too excited and maybe too idealistic and just trying to do u everything like completely decentralized um for them decentralized um all the data somehow split between different I have said some architectures that made no sense and just sounded very very nice in paper because they they just took everything to the extreme but it it just is not very good as an idea of practical. So um as I also said in u in proto protober yesterday yes yesterday before yesterday like we shouldn't let the perfect become the enemy of the good like decentralization and privacy they are not like binary it's a scale and probably you should go step by step and go more towards the privacy aspect and the decentralization aspect in your application but go step by step you can't just build everything at the right was part of the scale from the very beginning. >> Yeah. Yeah. Makes sense. Makes sense. And then there is another thing like how to stay sustainable, right? And you have some certain experience of like uh public goods and fundraising. So maybe you can share that because it's not always obvious especially when you're doing like selfish private uh not obvious like where to get you don't have access to this major VC money. How do you survive? >> Yeah. So there is where I I would like some advice from someone like it's it's it's not easy. I I I I am in cheat mode because I I believe I have cheated here since I have um been on the field for so long that people know me very well. So my name car some weight and this may give me probably does some advantage in all of these. the the funding programs and everything. People want to help because they know me. So maybe try to build a name for yourself first. But maybe this is not very a very good idea. Maybe the other one is try to find funding. Like there is funding in funding. There is VCs that are bad. There is probably VCs that are good. And maybe you don't need to raise a VC round. You can really just do a small Cindy round. Uh there are so many agents out there especially in crypto people who made their money early enough and want to um just uh help you know give back. So this kind of vain people that they their purpose is not to do 100x on their money they just want to see the the field flourish. Um this is a good uh option to just try to go the uh funding ground but not through the bad VC that will basically tell you no just turn this all into a centralized app and you know put your users in a world garden world garden then uh at some point you will see about monetization. Second would be monetization. try to monetize the product. Not too early because probably that will uh scare away people, but think of how you can make money. If you're making something that will never make money, um then you should think on a totally different direction. You should think, okay, we we need a we should have a charity foundation trying to always leave out of of of public um grants and funding. This is not something that I would advise for. It's very difficult and this money will probably run out at some point. to have. Yes, it's a quite useful advice because it's I mean obviously not easy to be a builder especially like if you just start this journey right and you don't don't have like name connection whatever uh like yeah what should you do but then um I don't I don't know um I do you want to talk about like maybe challenges and problems in the intersection of like privacy and the ethereum ecosystems like what do you see is broken now and what can be fixed like in in the nearest future but not going too deep technical. >> Okay. Yeah. uh I mean for for privacy uh in Ethereum and transparency I guess I guess it's this these two um uh attributes that are kind of confused like we use Ethereum a lot for for transparency because everything is public like most Ethereum all the public blockchains but Ethereum especially um because the feel that I know the best we like to use it and advertise that it's transparent and it can help against corruption let's all use the chain so that everything is is traceable But of course this creates a lot of problems for the individual like the privacy in Ethereum is basically not a register that there is no way to have privacy in Ethereum today and I mean we all know tornado cast and uh what attach any privacy protocols have had in uh in Ethereum. So I I think that what we would need to do as a as a as a system is to as a ecosystem. >> Mhm. >> Is to push more for accepting privacy like uh maybe have u organizations like the EUCI to help push regulators to understand that privacy is a requirement. uh because if we continue this path uh there is no privacy. There's not going to be any privacy in public blockchains and with the criminalization of privacy protocols then it's really not advisable to be using the public blockchains anymore. Um because it's just too dangerous to do something and that something be there forever. Like it's not just funding like understanding where funding comes from where money moves but it's also about actions. So like uh voting a specific way if the DA vote or donating to a cause let's say I don't know the Ukraine um war or something like that and then there is someone who sees you sees that you donated to a cause that maybe they don't like and that's there forever and like it's trivial to for anybody to see. >> Yeah. Yeah. Yeah. >> 100%. And it's it's very controversial. I mean that sense and I would really like to go there a lot because like that's what we do. We work a lot as web personnel. We work a lot with policy makers with lawyers and we try to advocate for that and make regulations also more uh comfortable for everyone in the market >> and it's a really good topic to explore but we have few minutes left so let's go uh a bit practical and I would ask you for advice for someone who is just exploring the world of privacy like on the 101 level like how could someone like what three habits let's say someone could start practicing a person who is not private but is interested in and like what are three habits that a person can use or integrate in his or her life to kind of be more private. >> Mhm. Mhm. Uh go to Google Maps and disable this tracking thing that they had. >> This is number one. Um second of course like use encrypted communication not not telegram not WhatsApp I would say but signal probably probably there is others um keybase I think is also pretty safe u and then um if you're into crypto um separate stuff this is extremely important even for um uh OG's they make this mistake very often um understand that this uh like you have one or two addresses. They become your public addresses and they are your public addresses. You kind of silo them. You you make them they only talk to each other and to exchanges. Nobody else. You have to be extremely careful about that. And then you have your private addresses. They should never talk to each other because then this this link will be there forever. Um yeah, three. I think this is three. >> Yeah. Yeah. Yeah. That's good. That's good. That's super good because like many people they count to space and they're like, "Okay, what should I do? There are so many like agreed I understand but like practically how do you want to implement this like privacy way of life but uh I want you to reflect a bit on the future. So imagine we are like 30 years from now and we have the internet that we dream of like private secure u whatever you wish it to be like uh how do we imagine this internet and how do you imagine life like if like all our dreams are true. there. >> Yeah. Uh that's a very hard question to me. I I have a very bad imagination for this stuff because it all goes badly. >> Yeah, I know. That's why I'm asking it >> because there is so so much negativity now. The world is quite So >> yeah, >> but still we are doing a lot to kind of move forward and I hope eventually it will move us somewhere. Maybe that's why I'm asking not like tomorrow but in 30 years >> somehow something could switch us to using uh using local machines more I'm very local maxi especially >> and I have lost a lot of um like I have a lot of despair with the AIS and LLM because I use them they're so useful >> and contrasting a local LLM right now with like the best models that I can use by paying open AI or anropics or cl or GPT they're so difference that we need to we need to make those uh better like close. So if everything so if we have a a world where we can um transact privately for everything like we as individuals that everything is private and I can myself provide um uh proof of this transactions to a third party without actually revealing anything else. So some protocols that are already being um explored right now. Um so for transaction privacy it would be pretty pretty cool and then uh for everything else that is government side to be um uh public like this is what we wanted public for right somehow we have become completely transparent while most of the governments we are still wondering what the hell are they doing >> and abolition of KYC I just don't want KYC to exist that would be great some kind of I I actually talked with uh the founder of Idos and he was basically kind of trying to explain that okay KYC is basically identity right so we need an identity protocol that's self-s sovereign that that will be very very good so that you again selectively can just disclose to someone your identity but not say is the whole world a leak here and there like is happening right now from the privacy perspective that would be pretty cool I hope it happens faster than 30 years. But with the LLMs and stuff, I'm the timelines have gone a bit crazy. Like we're talking about ADI and everything being done by LLM and like they're talking about year. So 30 years I don't know. >> Oh, okay. One more question like a hot take on AIM. So uh the thing is it's it's so big now, right? AI was web two, web three, whatever. and like everyone is using it and doing it very unconsciously and like not very secure and private. So maybe you can recommend something like how do you use AI if you ever should use it but still not doxing yourself too much still staying private how to do it like it's a secure way right >> I mean the only secure way is to run a local model through lama like it's it's not very difficult uh it's actually pretty easy like they have very good guys out there you can just do all lama and run various models the deepse that's an easy one >> was the one that is is is probably the best, but you need computing power. It's basically it's a hardware problem. So, I use them. Yeah. And but I I they're not private and I don't know what to solve here because it's not just convenience, it's worse. Uh it's productivity. This is not anymore about convenience. And we're talking about a huge difference between I have a beefy desktop. it should be able to run well but it doesn't compare to the latest cloud or chat GPT and that's bad because I have to also for my business right like if I can tell cloud cloud code to write code for me uh and solve three issues at the same time while I'm making my presentation for DO for example this is what I'm actually doing right now um then I will do it and there is there is no choice. If I don't, I my product will be irrelevant. I I'm pretty sure that we need to solve this. This will become probably the biggest privacy question in the next years. How can we make LLM private and I don't think so the thing that you would think okay local but then we need to somehow level the playing field because the local models versus the models that anthropics has and you have to pay for and god knows what they have internally I don't want to imagine what they have internally it's so huge that I don't know how we can close this gap we probably need a way to I don't know pull resources between some kind of open AI I oh wait open AI an actual open AI that u is is open and then they all put resources to create an open model that's actually better would be super great but at least as good as the ones that they have. >> Mhm. >> Because otherwise we we will have to use them. >> I don't think there's a choice anymore. If you don't use them you are going to be left behind. >> Yeah. Yeah. And that's first it's really I really like your thing about like productivity because it's actually productivity. >> It's insane. I I as as a developer I think that I I was so I I really love development. I love coding myself but it it does so much and so much faster that it makes mistakes sure but maybe it will become better that it will even not make those mistakes and maybe it will become better than me in a few years. So it will become a must that you just have to use. It's not a choice anymore. And that's why we have to go there before this happens. >> Okay. And then the last one because I really don't want to hold you anymore. Uh like quick recommendations. We are doing this once again for the academy. So maybe you can recommend some like cultural piece of maybe a book, maybe a podcast, maybe a film, something that you think could expand like anyone's like I don't know level of perception. It might be autism like how to enrich yourself, right? >> And okay, that's hard. That's really hard. Yeah. >> I don't know. Lately, I've been listening to to podcasts. I mean podcasts are good like I >> for example AI stuff and where we are going like I really love the hard podcast that it's it's really a lot about tech and they really do a big focus on on on AI and then uh >> yeah I I I don't think I have any >> it's fine it's fine like podcast is perfect >> yeah it's a >> probably more but I I I have to write them I just some don't remember >> no it's fine it's fine listen up for like super quick interviews that we have. >> Yeah. >> Yeah. Maybe we'll catch up at Dubcon and we will have more time. >> Yeah. Maybe. Maybe. >> Yeah. >> Okay. So, thank you for the interview. >> My pleasure. >> Yeah. Uh, ciao. [Music]