Karel Kyovsky on Manufacturing Bitcoin ATMs

WBD136 - Interview with Karel Kyovsky (Banner).png

Where to find the show

iTunes | Google | Spotify | Stitcher | SoundCloud | YouTube | Deezer | TuneIn | RSS Feed

Download Episode MP3 File
The file will open in a new window. Click down arrow to download the file.

Location: Skype
Date: Sunday, 21st July
Project: General Bytes
Role: Owner

Since the first Bitcoin ATMs started appearing in 2013, the market has grown at a significant rate, seemingly unaffected by Bitcoin’s volatility. With a fast-paced market and strong competition, manufacturers have to keep up with a growing demand for machines and new feature requests.

General Bytes has become the largest manufacturers of Bitcoin ATMs with over 1,700 machines in over 40 different countries (according to Coin ATM Radar). They are now faced with navigating the regulatory landscape and finding a business model that is sustainable in this new and emerging market.

In this episode, I speak to Karel Kyovsky the owner of General Bytes. We discuss how they stay ahead of their competition,  the challenges of implementing the Lightning Network, security and regulations.

TIMESTAMPS

00:03:04: Introductions
00:03:36: Background to Karel and how he discovered Bitcoin
00:06:01: Discussing the beginnings of General Bytes and their first commercial ATMs
00:09:02: Delving into some of the biggest challenges, including regulations and operations
00:12:09: Discussing Karel’s competition in the space and the differences between the companies
00:18:30: Who is responsible for adhering to regulations and licenses, the manufacturer or operator?
00:21:01: Exploring the ATM software and discussing what coins can be sold on the ATMs
00:26:05: Delving into the pros and cons of open/closed source software in ATM manufacturing
00:29:25: Discussing hacking and other attack vectors that Karel sees with his ATM machines
00:36:17: Touching on how prominent bad actors are in the Bitcoin ATM space 
00:38:36: Exploring what ongoing support General Bytes offer customers, after purchasing the ATMs
00:41:17: Discussing the economics behind the ATMs and the recent price increase at General Bytes
00:45:42: The potential of buying Lightning Bitcoin at the ATMs and the challenges in setting this up
00:51:52: Touching on other challenges in operating a Bitcoin ATM business
00:52:05: Discussing if increased competition from new ATM manufacturers is a good thing
00:56:03: Final comments and how to stay in touch

 

SUPPORT THE SHOW

If you enjoy The What Bitcoin Did Podcast you can help support the show my doing the following:

If you are interested in sponsoring the show, you can read more about that here or please feel free to drop me an email to discuss options.

SPONSORS

THANKS

A big thanks to my WBD Maximalist Patrons for helping support the show: JP Petit, Logan Shultz, Seb Walhain, Steve Foster, Tony, Gordon Gould, David Burlington, Jesse Powell, Bitcoin Tina, BitHyve and Wiel Menger.

TRANSCRIPTION

Peter McCormack: Hi there Karel, how are you?

Karel Kyovsky: Hi Peter, I'm fine. Thank you for inviting me!

Peter McCormack: No problem at all! So as you're aware, I'm going to be doing a special on ATMs. I've already spoke to Gil over at Athena. So I've learned a little bit about the operators, but now I want to learn a bit more about manufacturers. This is entirely new to me. This is an entirely new part of Bitcoin to me in crypto.

So I'm kind of at your mercy to learn about this, but before we get going, it's always nice to find out the background of the guest. So could you just give me a bit of your background and how you came into the whole Bitcoin thing?

Karel Kyovsky: Sure, so my background is the programming. I was programming since I was 10 and when I sold my first company in 2011, I moved my office next to the Hackerspace here in Prague in Czech Republic and the guys were there mining Bitcoins at the time and I thought it was ridiculous. Also I was working there with one friend on the token for the bank and then he moved to Trezor.

When I was doing a project for my new company, General Bytes, before I didn't know that I will be doing ATMs and I was working on embedded android computer. I wanted to do the embedded computer and I did the Indiegogo campaign and it completely failed. It was a big failure! I was looking for a project where I can demonstrate that it's a good approach to have the android in the devices, in the appliances.

I was contacting the heating systems manufacturers and they were not interested. They were interested only to have the cheapest device possible and they were not interested in the connectivity to the network. Because I was influenced by the guys in the Hackerspace, I thought, "okay, maybe I will do this proof of concept, I'll create the Bitcoin ATM.

I already saw Lamassu and I saw the Robocoin machines in the news and I thought, "yeah, maybe I can do this better and I can just show it off." After I think three months, I had the first machine, which I put into the Hackerspace and ever since then I and other guys in the company, we were just improving on the product until it's finished. It took us six years and it's still not perfect.

Peter McCormack: Six years!

Karel Kyovsky: Yeah, it's from 2013, we actually started.

Peter McCormack: When were your first machines actually available for the public to use? Was it 2014?

Karel Kyovsky: Yes, it was February, 2014 and that was the first installation in the Hackerspace. It was there also for hackers to play with that and they were actually able to hack the machine. So the first bounty I actually paid, was to the hackers in the Hackerspace. They were playing Angry Birds on the machine! So that was the very first security audit of the machine. But after that, we improved a lot!

Peter McCormack: When was the first commercial machine, like we would see today?

Karel Kyovsky: The first commercial... And after that, I think it was May 2014, there was a conference in Amsterdam, it was called Bitcoin 2014 and that was I think the first big show there, the first exhibition. So I went there with the machines and there I presented the machine. At first, actually the guy who was exhibiting next to me, he bought the first two machines from me.

So that was the first sale, but I delivered one month after. He wanted to take the machines from me directly from the booth, but I told him, "no, no, no, it's not finished!" That guy, the very first customer, his name is Martijn Wismeijer and he is now working for General Bytes as a marketing manager.

Peter McCormack: So there's the connection!

Karel Kyovsky: Yes, so the first customer is actually also a shareholder of General Bytes.

Peter McCormack: Were you the first company to have a commercial Bitcoin ATM in the market?

Karel Kyovsky: No, not at all. The first machine was I believe, developed by Lamassu. They had it as a part of some hackathon. It wasn't for sales, it wasn't for commercial purposes. So then was the Robocoin machine and I think they launched in 2013, somewhere in November, something like this, if I'm not mistaken.

They were the first who were selling the machines, but they were delivering a year after or half a year after that. There was a big demand, so nobody expected that. So the first I think, commercial available was Robocoin machine.

Peter McCormack: Were you an operator as well as a manufacturer or were you only ever a manufacturer selling them onto others?

Karel Kyovsky: We do operate the machines in Czech Republic, in Prague, but only for the purpose that we know what it takes to operate the machine, but it's not just a theoretical project, but that we also know what are the pitfalls, what makes it more effective to operate the machine.

Peter McCormack: Okay. Talk me through the biggest challenges that you have, both technically in terms of regulations, in terms of operations, in getting a machine into the market, because it can't have been easy?

Karel Kyovsky: So I think the biggest challenge was to enter the US market. I think that was the first challenge, to convince US customers that they should try our product from eastern Europe. So the beginning it was hard because we were also number five. When we started, there were already four manufacturers ahead of us who already had the deployments.

So I think at that time it was hard. We had to talk hours with every customer and now when we have so many deployments and we have some track records, it's not so hard sell. So at that time that was hard. Also, our mail server once was hacked, our company mail server. So that was also hard for us, because really it was a surprise and we had to lock everything and find out what happened.

I was really scared and we didn't know how to communicate to that customers. So luckily for us, only the mail server was hacked at the time and we transparently communicated that to the customers and they were okay with that.

Peter McCormack: Did you have to go through any regulatory process to actually enter into the US market?

Karel Kyovsky: No, we didn't have to. The only thing which we went through is the certification of the device itself. But that's easy. We had to set up the office in Florida, which we did a couple of years after.

Also I tried to operate machines in US, in Florida, but that didn't go well, because I started and founded the company there with the guys from local community, Bitcoin community and because I spent all my time in Prague, basically they found out that they don't need me. They don't need General Bytes, that they are better alone.

Peter McCormack: That sounds like they almost reneged on a deal? They kind of agreed something and then said, "yeah, we don't need you. Let's get rid of you."

Karel Kyovsky: Everyone I asked, it's always like this. If you do the business with people and you are not physically present there. I was going there only once a year, just to do the accounting and I was helping here from Prague, just by sending machines, providing the loans and it didn't work out.

So since then I said, "okay, I'm not going to do any business like this. I'll do business only in the companies where I am physically present."

Peter McCormack: Okay. All right, well listen, when I was looking into the manufacturers, the two main ones that stood out were obviously yourself, General Bytes and Genesis Coin, but also, and you just mentioned them, Lamassu. I was also aware of those. Are you essentially the three main manufacturers in the market or are there lots of them?

Karel Kyovsky: I think, I think you also missed BitAccess, who were also one of the first manufacturers. Then somewhere like maybe a one year ago they had pretty low number of the machines, but then they joined with Coin Cloud I think and they have pretty high number of the machines as a single operator. So I would mention them too.

Peter McCormack: So what are the main differences between yourself, those guys, Lamassu and Genesis Coin? Say, if I wanted to be an operator and I was thinking, "right, these are the different manufacturers I could go to, these are who I could order from", what are going to be the main differences?

Karel Kyovsky: Okay. So it's hard to talk publicly about about the competition, because I might not be correct, but the main difference is the Genesis Coin, that's number two now, but basically they have the same market share as we have. So they don't manufacture the machines. They have third party, which is manufacturing the ATMs and they just develop the software. Basically they send the CD to the factory and you get the machine and they charge you like...

Peter McCormack: It's like 1-1.5%.

Karel Kyovsky: Yeah, they charge you the percentage on top of your exchange rate. So their business model is basically like licensed loyalty fee. They are selling the big machines. I think their price is starting around $6,000 or $7,000. We went with a different approach. Maybe I should also mention Lamassu.

They are manufacturing very nice machines and they were the first ones and they are really focused on the design, on the UI. But their machines are expensive and they made the software open source. But the open source software doesn't work much in this ATM market because no operator wants to contribute the software to the shared repository.

When they develop some features, they want to keep it for themselves and they aren't interested in sharing the competitive advantage. So I think the Lamassu for example, disadvantages that the software is not that good, but it's open source. Also I think they are not contributing that much, they don't write that much software as we do, in terms of, for example, source code lines, if you will.

Peter McCormack: I guess the primary role of the software is the relationship between the user and some kind of exchange for either buying or selling your Bitcoin?

Karel Kyovsky: Yes, the software is actually the essential part of the ATM operation. Usually the users see just the physical machine, the metal and the computer. But the main thing is actually happening on the back end, on the server side. So where the transactions are being executed.

So yes, in some scenarios it can work that you insert the cash into the machine and server will buy on demand on some exchange the coins and send them to the customer. In some regulations in US, for example, you have to sell the coins from your stock, from your wallet.

So in the wallet management, the connectors on different exchanges, the different strategies, how you handle the coins, the remote configuration of the machines, the stability of the software that the machines are 24/7 operational, that's important. I didn't finish on the competitive differences if I may.

Peter McCormack: Yeah, I need to know yours!

Karel Kyovsky: So we went with the approach to have the multiple different models. Right now we have, I think five different models. We have also the small, the cheapest ones. So I think that's the reason why we were successful, because we had the very cheap model, but customers who are interested in more secure solutions, can buy the very expensive one, well not very expensive but more expensive model.

So it was I think the range of the machines, that's one of the differences. The second difference is that we really invested all the money which we made on the machines. We invested into the software, so we have quite big development team and we work everyday on improving the software. So far we didn't have the service fee on the ATMs, like Genesis Coin has.

But we will have to introduce that and we will signal that to the customers that we will have to start charging several license fee, because no longer is it possible to finance the software development and the maintenance of the software and the new features, from the sales of the machines.

Because there is a lower and lower number of the new customers, new ATM operators entering the market, due to higher regulation barrier, let's say. It's hard to find a bank which will support you in operating the machines and also it's hard to get all the licenses which pop up in every country around the world.

Peter McCormack: Who takes responsibility for that? So if I was to order a machine, say from you, am I buying a box with software and I have to do everything else in terms of regulations and licenses or do you arrange for the licenses?

Karel Kyovsky: The operator is responsible for all of that. We provide you with the sophisticated tool, which you can configure based on your AML policies. You can configure it, that you want to collect the phone number from the customers, scan all the ID card, social security number, all that and it's your responsibility. 

Basically, you as ATM operator, you have to build the KYC/AML policy and you have to find the lawyer in your area to help you to identify what your responsibility is in your state, because in the US, it's different state to state and it's impossible to guarantee anything for the customer. It's changing everywhere, every year, in every country!

Peter McCormack: So you essentially provide the box with the software for them to set up, based on their own country and their own regulations, which is great. But also, I guess what you're saying is, it's very difficult for you to fund and operate a business based on just machine sales alone, because I guess it's an inconsistent model. You need the consistency of a software license, so you can manage and update and support the software?

Karel Kyovsky: Yes, I think the market... Before it was possible because the market was still immature and the new operators, not the ATMs, but the operators were popping up every week. But I think now when the market where the rules are tougher and tougher, the market consolidates.

There's one operator buying another operator. So it's no longer possible to grow that fast from the machine sales. But the cryptocurrency ecosystem is still involving, for example, Monero and other cryptocurrencies, they are still adding new and new features and somebody has to add the features to the ATMs to handle those. So that's why ongoing payments are needed.

Peter McCormack: So you can buy Monero from one of your ATMs then?

Karel Kyovsky: Of course.

Peter McCormack: Okay, that's cool! So what coins do your machines support? Can it support anything or are there a certain number of restricted coins it can support?

Karel Kyovsky: Okay, so first of all, we support some basic set of the coins, but you as an ATM owner, you can program support for your own coin. So we have a GitHub repository, which is open source and they're examples of how to implement support for your coin or for different coins. So for example, Bitcoin Cash is there or I don't know, something else.

So of course we support Bitcoin, Litecoin, Ethereum, Dash, Bitcoin cash, Monero, ERC-20 tokens, but some of the currencies we don't support two way because that's more complicated. So either the customer didn't implement it in the GitHub repository and is still waiting for the contributor to add his contribution there or there was no demand or it will be added in the future. 

Also recently we added the Lightning network support. I'm really curious how the Lightning network will work on the ATMs, because it has its shortcomings and I'm interested in how it turns out and if it will be used enough by the customers.

Peter McCormack: Okay, well we'll come to Lightning, because I'd like to talk to you about that in a bit more detail. Let's talk about the fact that with your software, that people are able to add their own coins. When you said there's not two-way support, what did you mean by that?

Karel Kyovsky: One way support to be called the direction from cash to crypto, which is essentially easier because basically you accept the cash and you just send some coins from wallet to the destination address. The two-way support means that it's cash to crypto but also crypto to cash and that's more complicated because you have to tell the customer where to send the coins.

Then you have to check if he sent enough, if he's sent enough fees, if the transaction has enough confirmations, that differs based on the cryptocurrencies and after that, you let the customer to withdraw the cash from the ATM. So that's more complicated and two ways is both directions.

Peter McCormack: Has that received any kind of negative feedback, in that you're allowing for support of lots of different coins, because I guess different coins have different kinds of respect within industry? Bitcoin is king and as we know, lots of other coins are known as shit coins. Has that led to maybe just a lower quality of coins available on the machines?

Karel Kyovsky: So we as a general bias, we implement support only by ourselves, only for coins, which we think are valuable. We encourage customers, they can implement any coins they want. We don't care because we are not here to judge. I personally, I am not a Bitcoin maximalist. I think there should be competition between the coins itself.

I believe the coins are basically the applications and it's healthy to have a ecosystem where you have multiple coins and they steal the features from one from each other. I think without the Litecoin, we wouldn't have the SegWit yet and only I think because the Litecoin adopted SegWit, then it pushed Bitcoin miners to use the SegWit or to approve the SegWit too.

So yes, we had the incidents where we saw that the customer wants to implement some basically scam coin and we told them that we will not do that. If they buy machine, we don't care, but we don't encourage that.

Peter McCormack: So there is a risk that someone could exploit the software to promote a scam coin and you guys can't control that?

Karel Kyovsky: It's not exploiting because basically every operator has to enable that coin by himself. It's ultimately the decision of the operator to enable that support and configure the support for that coin, because the operator has to install and configure the wallet for that coin etc. So we are not here to judge the operators and what coins they sell. I mean if somebody creates his own coin and then he buys the machine, why not?

Peter McCormack: Okay, so it seems kind of interesting how some of the manufacturers have gone for kind of open source and some have gone for kind of closed. What do you see kind of the pros and cons of both approaches?

Karel Kyovsky: I think for the open source, you have to have the other business model. I think it's really hard to find the business model with the open source. There are some products which are good for the open source and some are not good for open source. Typically I think the open source should be used for the infrastructure, but you have to have the critical mass of the users. So if you have open source mobile phone, it's okay because you have so many developers being able to hold the mobile and program for it.

With the ATM, you have only a little of ATMs and it's really hard to find developer and tell him, "hey look, program for this ATM." So I think the open source doesn't work on this. The advantage is that you can tweak anything on it, almost by yourself. You can tailor it exactly to your needs. You don't have to care about the other features. The closed source model, which we have, we have some parts open source, some part we have closed source.

You have open source only the part which enables operators to program plugins. We call it extension for the server and the support for the new coins. But for example, the ATM software is closed source and the core of the server is also closed source. The advantage of the closed source is that you have it under control. So you can really spend time time on perfecting those features.

Yes, you have to invest in that feature and you have to basically write features for all of your operators, as they can't write it themselves. But they get it as a package and it's well tested. We can guarantee that this release has been tested and with open source software, you don't get so many developers. With the closed source, you might be able to have more developers because they have incentive.

But there should be other advantages but I can't think of it. It's more integrated with the hardware, that's also one of the advantages. That we can write the software for just a few models which we have, we don't have to write the software which works with 10 different cameras, 10 different bill acceptors, so it helps on this side.

But the open source software doesn't work I think for the ATMs, because ultimately on the end, the user is ATM operator and that's the business and not the consumer. The businesses are interested in something which is working, which may cost money, but they also don't need it for free because it's generating their profit.

Peter McCormack: Yeah, exactly. Interestingly, today I released an interview with Nick Percoco who's the Chief Security Officer at Kraken. So I spent a good hour and a half with him, learning about the history of hacking and all the main challenges they have with the exchange.

I imagine hacking and security is quite important in this industry, not knowing what the attack vectors are, but do people directly try and hack into the machines in locations or are they remotely trying to hack? What are the attack vectors you guys face, as a manufacturer and how important is software and security with what you do?

Karel Kyovsky: So I think in terms of software, I mean, no software is perfect. So the only thing which we can do, is to have regular security audits. So every year we give our software to a security company and they to hack it. Not the real production system, but the test system with exactly the same software. So they try and they check if we checked properly the certificates, etc.

So that's the software. We also get from some of the customers their security audits or results. So we had customers who provided us with the reports from their audits and we implemented their findings or fixes for their findings. The main attack vector of the ATMs at this moment is the physical attack. So if you have a 25kg ATM and you take 50kg a hammer, it's going to be the hammer, which is going to win.

So we did lots of improvements on the machines, that's also good part of having the design under the control that we made, with every batch we add new feature. Not add, but to change some of the things on the machines for example, to assemble it faster or to manufacturer it faster, but also to improve the security. So we did quite a lot of security checks.

We also did the crash tests on the ATMs, to check if somebody hits it, where it's going to bend and how to bend it and still still have it closed. So attack vector is the hammer still.

Peter McCormack: So that is people what, trying to get into the machine and steal the cash?

Karel Kyovsky: Yes.

Peter McCormack: Okay, but is there any other attack vector where people are trying to hack the machine to trick it into thinking money's deposited and buy Bitcoin? Is that any kind of attack you face?

Karel Kyovsky: I noticed, I think it was on the beginning of this year, there was an operator who was attacked by the zero confirmations attack. Basically somebody wants to withdraw the cash from the ATM and the operator enabled the zero confirmations. So zero confirmation says basically that you give somebody the cash before the transaction gets into the Blockchain, which we don't recommend of course.

But these algorithms work, that for example, you watch the Blockchain, you calculate how big is the fee, what is the chance to get the transaction into the block in the reasonable time and based on that, you can calculate the risk and you can allow this strategy. Or for example, if you see competing transaction in the memory pool, you will disable the withdraw.

But still this is the only software attack which we saw so far and this is something which cannot be fixed. I mean on Lightning can fix it or the Dash instant send transactions can fix this, so that you can instantly be sure that the coins are yours. All other operators work around to suit their frameworks to calculate the risk.

But always, if you transmit the transaction with high fee, you can always double spend anything, it's just matter of miner to accept your transaction. You can do anything with that. Once we saw on the Wall Street market that somebody was selling the exploit for Bitcoin ATMs, but it was a scam, it wasn't real. So other attacks would be the DDOS on the operator, but mostly the attacks are just the physical attacks.

Peter McCormack: Interestingly, this part of the Bitcoin industry is still kind of relatively unknown. It's kind of like... So for example, if you want to go and use an exchange, pretty much everybody knows all the major exchanges. I could tell you the difference between Coinbase and Kraken, the pros and cons of both and most people in the industry kind of know this, but within the ATM industry, I guess it's less well known.

Is there the opportunity for, in the UK we would call them "cowboys", like "cowboy" manufacturers to come in, create machines, sell a dream and then actually not be able to deliver? Is that something you've had to kind of combat as an industry?

Karel Kyovsky: That was happening in 2014 I think. These small startups which created few machines and then they raised money and then they went for different projects for example, I think that was a BTCPoint, which I think was a Spanish ATM manufacturer.

In 2014 they were exhibiting in May and I think on the end of the year, they were already in San Francisco working on different projects and there were lots of companies which were working on the ATMs at that early time in Asia too. I think there were at least 20, 30 different companies which tried it.

But I think only the ones who are just just keeping on working on it survived or were successful. I think it was just a matter of not taking other opportunities but just work on it.

Peter McCormack: Okay, so if I was considering moving into this Karel and I was saying "right, I want to get a Bitcoin ATM in Bedford. I want to work with General Bytes." What kind of support do you offer to somebody, firstly, as somebody new entering?

So as somebody completely new, buy my first machine, what kind of support do you offer in terms of setting it up and getting going? Then what kind of ongoing support do you offer companies who kind of want to expand and say, after buying my first one, I wanted to buy a few more. How does that work?

Karel Kyovsky: Right, we would recommend you to order the machine through our website, that's the channel which is streamlined for you. All the questions you have, sales will answer and when we sell you the machines, we send you the machine via DHL, you receive the machine, you write us that you received the machine and we give you the keys for the software and you are ready to go.

In the manual you will find the links to YouTube videos on how to configure the machine. We are constantly redoing the videos and basically those are hours and hours to watch what to configure or how to configure it. There are also videos on how to order the machine. Those are your first steps to learn how to configure the machine from the YouTube videos.

We also had the support guys here in Prague and in US, where if you are a customer, you can send their email and we respond to you in 24 hours. So I think most of the questions we get is, "how do I configure this?" And second is "can you please add this essential feature for me." Also, we received the software box back reports from customers, which we try to fix with the software releases, which we do every 14 days.

Peter McCormack: So if somebody requests a new feature, is it something you have to provide for everyone or do you go down that route of doing customized versions for each customer?

Karel Kyovsky: We try to deliver the feature for everyone, like to implement it free of charge and deliver it to everyone, because otherwise it wouldn't make sense for us because we have limited resources. We have only a few programmers, which work on the ATM itself. We have programmers for server, programmers for [Inaudible 40:31] and we just can't do the custom development for everyone, we don't have enough developers.

So yes for some cases, we do, for example, the software audits. When a customer writes his own extension of the server and he sends it to us for audit, yes, we can do that, that's quick. But mostly we tell them to write an improvement ticket and we will add it into the pipeline, into the roadmap and based on how the feature is complicated or how important it is or how it fits existing roadmap milestones, we will implement it.

Peter McCormack: Okay. I'd like to understand a little bit more about the economics behind this and I've done my best to do my research and kind of understand how you differ from other companies. As I understand it, up until recently, you had one of the most affordable solutions. So am I right in thinking your machines were around $3,000?

Karel Kyovsky: Yes, the cheapest model.

Peter McCormack: Yeah, the cheapest one and some of your competitors had kind of, you would probably argue more solid machines that may be $6,000 to $7,000 per unit. What's your kind of model now? What's your pricing model and what are the economics behind this? Because you did announce a price change recently, right?

Karel Kyovsky: Yes, so basically what we need to do, is we need to double the size of the software development team. So far we were able to finance that from the machine sales. The machine sales are decreasing basically. Not with every operator, but some operators are super successful, they buy 10 machines a week.

But because the global, and not just in the US, but globally the market is consolidating, just the sales sales of the machines are sufficient to finance the maintenance of the software and slow development. But we need to add new features and innovate. So we need a second income to finance that. So the plan is to double the size of the development team.

Peter McCormack: So that price change, is that only for new customers and new machines ordered or does that actually affect previous orders?

Karel Kyovsky: It's affecting everyone, every customer. Basically the customers, they can continue on operating their own machines, nothing will happen. But if they want access to a new software, new developed software, they need to participate on that.

Peter McCormack: Have you had much feedback on this price change?

Karel Kyovsky: I did get feedback. Actually I also established a new email address, so our customers could send me their arguments! Some of the customers, I gave discount for example for some limited time, so they can accommodate some customers.

They don't know the economics behind how expensive the programmers are nowadays, because even if you are sitting in Czech Republic, very far from San Francisco, you are still competing with those San Francisco companies, because they are able to hire the programmers here and because the video conferences, including our conversation, are really easy to make across the ocean. So they also have the American salaries.

So it's really an expensive exercise here too! So I think a majority of the customers, they accepted it. The biggest customers I visited personally, the ones who have more than a hundred machines. Some of the customers, they didn't like it. Some of the customers who are not satisfied at all, they had, for example, just one machine.

Peter McCormack: That's typical though! I remember when I had my advertising agency, it was the lowest spending clients that were the most demanding. You always tend to find that. I think that's in almost every industry.

Karel Kyovsky: Yeah, I think it has something to do, like if you have lots of machine, you start to understand and you have stuff you start to understand, that things cost money. So if you do ice cream and the water is more expensive, you have to increase price of the ice cream. But you just can't start to complain about the price of the water!

Peter McCormack: Well, you've got different economies of scale. If you're operating one machine, you're probably doing so on a very fine margin and any difference to that is kind of scary. Whereas if you can buy 10, 15 machines, you've probably got better economies of scale. So that doesn't surprise me.

All right, let's get onto the Lightning network, because you've mentioned support for that. I'd be very interested to hear how you intend to deliver it, what your thoughts are on it, what are the challenges that come with it? Because I think this is super interesting,

Karel Kyovsky: It's very early on to give a comment on that, but we have it on the ATM in a way that when you click on Bitcoin Lightning, you can open a channel optionally, or you can buy or sell coins. So there are issues. Number one is the capacity of the Lightning network. So not everyone is happy to lock in Bitcoins into the network just for the very low fees in the network.

So I think number one issue is the capacity. So for the small payments, like the coffee, it should work. If you go to the ATM, you will buy Lightning Bitcoin and then you pay for the coffee, it's going to work fine. But for the high values, it's not going work and the biggest issue is I think it's super complicated for the users. I think at this stage it's really... It took us six years to explain customers what is the Bitcoin address and now we have to explain them what is the channel?

That's super hard and you have to explain them, that the coins are locked in a channel, those are going to be released. If you send them somewhere, you have then zero, then if you want to send coins somewhere, you have to log them. It's really complicated. So I think the ultimate result of the Lightning network will be, that we will all have custodial wallets.

We will have the wallet in the mobile phone, but the node will be run somewhere else by a third party, who will be managing the channels. The users will not know anything about the channel. So it's going to be really long way away.

Peter McCormack: See it's funny, I totally agree with you. I don't think... How do I put this without pissing off some listeners, I don't think people are being as realistic as they should be about the current status of Lightning, which isn't meant in any way to discredit any of the good and hard work that's been done so far. I think the fact that if I wanted to send you an invoice, you could pay it and it would instantly settle pretty much and that's great.

But at the same time, the user experience is actually quite different from base chain Bitcoin, which itself as you said, is hard enough to teach people. To then teach this whole other experience, it complicates it further and I think people are avoiding this.

Karel Kyovsky: Yeah and Lightning essentially just leads to the centralization. It will have lower fees, but it will be more centralized. But from the user point of view, it's really bad user experience, but it's not the fault of the Lightning. I think it's just going to take time to standardize again everything and all these scenarios which might happen.

For example, this invoice, it has to be more flexible and we implemented the support for the Lightning on the ATMs, just to help to push the Lightning forward, so that we can find all the pitfalls and to help them move it forward and fix a few things. For example, the invoices, if you want to send somebody Bitcoins, he has to tell you first how many coins you have to send.

Peter McCormack: No, I don't understand that. I've always found that kind of annoying.

Karel Kyovsky: So, for example, in front of the ATM, if you want to send coins to the... No, if you want to receive coins from the ATM, usually on the base chain, basically what you would do, is you just go to the ATM and you insert the cash, show the address, that's it. But here with Lightning, he would have to first show how much Bitcoin you want to receive with the invoice.

Then insert the cash and then to receive it. So we had to use a few hacks, that it is an invoice, to doubt the amount. There are some scenarios which are not tested yet. So these are the things which work for example, with Eclair wallet or node or doesn't work with LDN. So, this needs more time and I'm worried that it's going to take again five years.

Peter McCormack: We don't want another five years! If it's another five years, the BCashers will be throwing a lot of mud at Lightning for that.

Karel Kyovsky: I don't know, but maybe there is no other choice.

Peter McCormack: Yeah, well we'll see. Look, I like it and I actually have started to use Lightning, but sometimes I'm actually comfortable spending a little bit extra paying the on-chain fee, to do an on-chain transaction, than to go through the hassle of doing a Lightning transaction because of the UX. It's like if I had to do like a $100 transaction and if it cost me another $1 or $2, do you do it on the base chain?

I probably would because Lightning doesn't offer me enough yet to switch over. But I just think that's because yes, we are still early, but give it some time. I'll probably get shot for that statement as well! All right, well listen, this has been great, I've learnt so much from you, thank you Karel. I've got a few more questions to close out with you first. So what are the main kind of difficulties and challenges for you in running such a business as this?

Karel Kyovsky: I think it's the regulation. I think now we have to add... I think we spent at least half of our time just for features which are related to the regulation. I'm thinking about the challenges.

Peter McCormack: You've got to think what keeps you up at night? What has you pulling your hair out?

Karel Kyovsky: I think the competition always, I think it's the competition. I think it's good always because sometimes I'm lazy to add some feature, but when you see that the competition has it or is working on it, just can't not do that! I think it's healthy, so when I see competition that they added something which we already have had for years, I have to come up with something else. So I think it's the pressure from the competition.

Peter McCormack: Interesting, so competition and regulations. If new manufacturers were thinking of coming into the market, obviously you would recommend they don't, because you don't want more competition! But if there are new ones coming into it, is that actually good? Is more competition good for you, if you're really honest?

Karel Kyovsky: Yeah, I think it's still good. I think it's good, if it's fair competition. If they play fair, for example, that they really deliver what they promise, because otherwise it's bad reputation for the business.

Peter McCormack: A bit like Robocoin?

Karel Kyovsky: Yeah Robocoin I completely understand those guys and why they failed, it's not their fault. They were victims of their success. They were not ready for that.

Peter McCormack: Because a business can fail from growing too quickly, you just run out of money.

Karel Kyovsky: I think they didn't run out of money, they just were not mature enough to handle all the demand. I think it was young team and a small team. They didn't listen to the customers, I think that's also one of the things which we really do. We don't add the features which we want. We add the features which customers want and I think that's essential too.

Peter McCormack: What about like what happened with Skyhook? I read about Skyhook, they had a very well priced machine and were selling to the hobbyist, but they ended up becoming quite unresponsive and poor at support. What do you kind of make out of what happened there?

Karel Kyovsky: My thoughts when I saw that, is that it's not sustainable model. Basically I think it wasn't generating any profit, so there was no incentive to continue in the project, that's number one. Number two, they were using bill acceptors, which supported only Euro and US Dollar and maybe I think not all the Euro banknotes. Maybe they also supported Australian Dollar, but that was all. Another point was that it wasn't all ready for the production use.

I mean it was still more a hobby than to use to create a network of it. The security was low, I think it was very small, but that was because it was cheap. The ATM companies are operators, they need a solid solution, which is always online and just pretty much secure and I think Skyhook didn't fulfill that. They were also using the open source model, which don't work for the ATM business or at least it didn't work so far.

Peter McCormack: Alright, well listen, this has been super useful. I feel like I've learnt a whole bunch more now from you. I don't think I'm ready to become a manufacturer myself, but maybe one day I'll buy myself a General Bytes machine, get a Bitcoin ATM in Bedford. But this has been really useful. Before we go, just tell people how they can find out more about your business, how they can get in touch, who you want to hear from, give a shout out for your company!

Karel Kyovsky: Sure, so you can read about us if you write into Google General Bytes and on our websites you will find links to our Twitter, our YouTube videos, etc. There you can find all about the ATMs and this is the product which we concentrate on. We don't do many products, this is our main product.

Peter McCormack: Fantastic! Okay, well look, it's been great to talk to you. Good luck with business and hopefully one day I'm going to get myself over to Prague, because I've never been there and we can get one of those good Czech beers in!

Karel Kyovsky: Sure, I will be happy to host you here! Thank you for inviting me.

Peter McCormack: No problem, take care.

PodcastPeter McCormackComment