Nextcloud: A Self-Hosting Alternative for All

Step into the Nextcloud HQ in Berlin as we are joined by Jos Poortvliet, Co-founder and Director of Communications for Nextcloud, for a wide-ranging and revealing look into the open-source, decentralized ethos of Nextcloud.
This in-person interview follows Jos's early days in tech and work at OwnCloud and sets the stage for a dive into Nextcloud's commitment to open-source, data sovereignty, and their broader mission to return data control to users as the company scales past 100 team members and thousands of customers.
Throughout the interview, we covered the business model of Nextcloud, learned about some lesser-known features and use cases of the Nextcoud platform (Hub 7), and learned how they integrate AI into the product while still maintaining their core values around open-source, data security, and privacy.

Ready to Set up Nextcloud?
How to Set up Nextcloud on Unraid
2023 Set up guide by WunderTech

Other Ways to Connect with the Uncast Show

• Uncast Show Website
• The Uncast Show on Youtube

Chapters

• 0:12 Getting to know Jos Poortvliet, NextCloud's Co-founder and Director of Communications
• 7:23 Nextcloud's origin story and the current state of the team
• 15:01 Free and Open-Source Software: What is it, how has it changed over time and how does Nextcloud work under this paradigm?
• 30:15 Nextcloud's features and use cases for home and business users
• 45:23 Nextcloud's implementation of AI into the product
• 54:45 Hardware recommendations for optimized Nextcloud use
• 1:02:35 Nextcloud's anti-competitive lawsuit against Microsoft
• 1:09:37 User Questions

Transcript

Ed: 0:12

Hi there, guys, and welcome to another episode of the Uncast Show. Now today I'm really excited to be over in the next cloud offices over in Berlin, and I've got a very special guest today. I've got Yosh, who is the marketing director of Next Cloud. I'm sure many of you guys have heard of Next Cloud. It's an absolutely awesome open source project that empowers organisations and individuals to take control of their data and everything to themselves. So, yosh, thank you very much for taking the time to speak to us. Thanks for coming.

Jos: 0:43

It's really cool that you came over all the way to rule a dark deng and slowly but live.

Ed: 0:48

It's been wonderful. Thank you very much for having me. Anyway, I'm sure our audience is eager to get to know all about Next Cloud, but I think, before that, myself and the audience get to know you a little bit better with a section I like to call Quick Fire Questions, and so the purpose of this is I'm going to ask you a question it's going to be an A or B answer and you're not allowed to think about it too much. Yosh, no kind of like taking 30 seconds to answer, or straight off the top of your head. Okay, android or iOS. Ios Text message in or phone call. Text and console gaming or PC gaming. Pc and what's your favorite game?

Jos: 1:30

Supreme Commander, that's it. It's a two life Forza, life's forever. And desktop, laptop, desktop. My love is there, with Linux or desktop.

Ed: 1:48

And what is your web browser of choice? Our folks, yeah, good choice. They're more Emacs. And what is your favorite open source project besides Nextbox, opensucha?

Jos: 2:04

And then I've reliable on Project System. I'm not entirely unbiased. We'll get to that.

Ed: 2:11

And hiking in the mountains or exploring a new city. A new city, yeah, especially if it's old, very old. Well, I think I'm going to know the answer to this next question coming up then. If you could travel in time, would you go back in time or forward in time, back, yeah.

Jos: 2:27

And I have the coffee or whatever the hell they drop back.

Ed: 2:30

then if I'm what I'll be or someone like that, there you are, so you've had an incredible journey in the open source world, from being active in KDE to serving as the open source community managing, and now you are needing marketing at Nextbox. Could you share a bit about your background and go into detail of what you were doing before you came into being a marketing developer at Nextbox?

Jos: 2:52

Yeah.

Jos: 2:53

So early 2000s I had the same problem as a lot of people had. They were on Windows 98. Windows ME was a mitigated disaster. And XP came out and it was terrible in the beginning and I decided to explore some other operating systems. I played with BEOS. That was kind of cool because you could play more than one MP3 file at the same time and it wouldn't stutter. That was still very special back then. Yeah. Then I settled on the next. I settled on Kabuntu React nopics. So that was one of these live CD things that are still fresh-ish. Back then, yeah, and I started using that and really enjoyed it and starting playing with it a lot. And then when visiting a tech show trade show-ish thing about tech back then you bought your CDs and DVDs still physically. At one of these tech shows there was a Kenny booth and they needed help and so I had read enough about Kenny to basically chime in. Then they asked, hey, can you come by the next week? We have a lot of refunds in Amsterdam. I was like, yeah, sure, and yeah, it went downhill from there basically. So I ended up helping organize KT marketing and I did that for a very long time until I got a job at SUSE as community manager for open SUSE and I did that for a number of years.

Jos: 4:22

A really good friend of mine, frank. He asked if I wanted to join him at Omblout, which he had founded. It was kind of cool because when he announced it in 2010, we were both at a KDE, a KDQ event in San Diego, even in the same hotel Epic Ties, lots of run and fun in general. And yeah, there he had announced that he wanted to create this self-hosted alternative to, back then, dropbox. Right, because at the time, dropbox and Google Drive was where more and more people were putting their data. And, of course, frank, he was at the board of KDEV at the time, both as KDE people. He saw a problem with KDE being great on a desktop and it was wonderful operating system with Linux and, as a desktop, kde on top. But if everybody puts their data at Dropbox and Google, then we actually haven't gained any computing freedom, right, like this was changing the game, basically. And so, yeah, in 2010, he basically announced he was going to try and change that.

Jos: 5:31

He started the project in January, I think, but the first code drop came in February or March and very quickly a bunch of people joined him. They already had their first meeting in about 2010 late in the year and in the meantime was working at OpenSUSE then as Community Manager and we kept in close contact. I actually often gave talks on OpenSUSE and on cloud on a Raspberry Pi. That became like a nice subject for events and we even had OpenSUSE people working on the first desktop client for on cloud. So there was a good relation between the projects. Sometimes we share the booth and stuff like that, and then at some point he asked me to just join on cloud. I was like well, you were already promoting in any way, why not just come on board and do that full time? So we became Community Manager at on cloud and well, when that fell apart, the company, I mean we started the next round together with a bunch of other people.

Ed: 6:32

How long has next slide actually been around in the form it is?

Jos: 6:36

June 2, 2016,. So we announced it June 1,. We started to build the website.

Ed: 6:44

And then, just going back to you know, same frame was worried about things like Dropbox and Google Drive. How long did Dropbox and Google Drive been around? So I can't remember when they actually, around 2005 fish that's the prime Been around for about five years at the time. Yeah, exactly Right. And so basically, next cloud was born out of only cloud, exactly, and I asked why was the fork?

Jos: 7:11

Well, so basically I went on bankrupt and obviously we had seen it coming. It wasn't going well. Yeah, management and complicated relation with investors and all this stuff and a couple which in principle I have nothing against. I mean it can be good, but you need to have good win-win set up, special with open source, and it can be tricky and you know, management kind of failed to to properly handle that. So, yeah, frank had already left in, I think, april or May and then, yeah, in June, basically I joined him and more joined in the coming two months while on crowds when bankrupt, and then what was left was bought up by a German real estate guy Right, who kind of kept it going a bit, and most of the engineers went with us and, yeah, well, now it's been sort of good.

Ed: 8:06

Yeah, what was that kind of unsorted?

Jos: 8:09

I knew them because they used to be called a salient. So they make like, yeah, also file sink and share software, Dropbox, like, and they're yeah, they're a bit bigger and then they went on shopping spree. They bought like a couple of file sink and share companies Drive through the May Tech, I believe. Yeah, so I'm guessing they're going to consolidate all their products and just make one product out of three.

Jos: 8:33

Yeah, I don't think they'll merge it. I mean you can't really. Yeah, I mean, what would you get out of that? I think just move everybody over to one of them. Well, I would be surprised if it is Dracul or, you know, one of the other small ones.

Ed: 8:48

So the team at Next Cloud how?

Jos: 8:50

big is your team? Well, a little over 100 people.

Ed: 8:53

Wow, yeah, and who are the key players?

Jos: 8:57

in the team. Yeah, I mean, many of the original people are still there not all, I think, and we'll have left three, four, five but there are people at the company several who have been involved since 2010,. Basically, yeah, I mean, it's kind of the thing of open source, I think, right, Like people care about what they do and yeah, I mean, if this project doesn't go bad, then you can keep doing it. So we have, like Robin he was a kitty contributor, I think, at the time and started to contribute to Next Cloud, initially doing a lot of the file storage handling. We wrote things like external storage feature and while he's still doing that, basically we've had Arthur. He comes from the Kubuntu community, so very close. He did LDAP back then. He recently moved on from LDAP to talk. He wanted to do something new after like 10 plus years, you know, 12 years.

Ed: 9:59

I've always found the external drive plug-in very useful. That's cool right Running an unread service. Basically, we can use that to connect to some of our normal shares yeah.

Jos: 10:12

I have to say I have an old server, like an old desktop I don't know, a server with a couple of drives. It has music and movie collection and private pictures and stuff and most of those are just local directories mounted as external storage and they also excellent directing via NFS. So yeah, I mean these feature, I mean there's something wrong. I know to talk to them. That's all you say.

Ed: 10:36

So when you mount some external storage, do you prefer to do it like over the network, like those in network external storage, or do you kind of mount those amount points into the doc container?

Jos: 10:50

So in my case, because it is all local, like I have one server and all the drives were in there and all their runs are now the next level in one Docker container and there I think when you think in the command line when you started, basically in the environment variables, you set like the amount points. So I did. But those are actually local drives. Yeah, and if there were network drives and I had them on the nest separately, I would probably use the native protocols or like FTP or Samba et cetera in next clouds, because it can then do smarter stuff, like a range find.

Jos: 11:32

If you go to your next clouds you want to watch a five gigabyte video, you want to start in the middle of the video. Well, next out has the ability in most cases to store them, to simply start in the middle of the file, right? So you want to first transfer the whole file and then see to the middle. That would be paid in the behind. And, yeah, I would guess that such features and support for these kind of things would come first to the native support. But also depends on how you've mounted it. I mean, I bet NFS supports this as well. So if you simply have an NFS mount and then you give next up access to that, it'll also be able to immediately go to the middle of the file. So I don't know how much difference it makes, honestly. But yeah, I'd probably still go for the native stuff, just to be sure.

Ed: 12:25

And so with the team, obviously we're in your office in Berlin and do you have multiple offices and is the team? What's the geographical location with the people?

Jos: 12:35

Well, I mean, we have people in Hawaii and Canada and India. So basically, I don't know, out of the 100 people or 100 plus people, I don't think more than 15% has office. The rest works somehow. That's still primary belt and I mean that's gotta stay that way, because we hire people wherever they are. So yeah, we have 10 people in Berlin, so that's technically our biggest office we have like seven or eight, is Stuttgart and that's it. So the rest works somehow.

Ed: 13:10

So you're a pretty modern company with a hybrid, but you know, yeah, absolutely. Oh well, companies started working remotely from COVID, but you've been doing it really since day one.

Jos: 13:21

But it's also kind of the open source model. You know, like this started as an open source project right back in 2010. And the nature of it hasn't fundamentally changed. Yeah, we know we have a company and you know you have as a laser of customers, like there is process and engineering and management on top, but all our work is still happening in GitHub. You know, people sometimes ask me for roadmap and I mean, what can I tell you? You can get up, because everything is there. There isn't. Yeah, it is all transparent.

Jos: 13:51

This is actually so transparent that a lot of people then tend to think that there must be more, so then they don't like it's really hard to get a people's headset. Yeah, what you see really is what you get. You know, of course we have like secret rebows because we have security issues, you know, and customer issues that we handle separately, but we really put in a lot of effort to make the defaults to be open. You know, like the chat rooms on talk on our company server, there are tons of community members in there. They're most of them are open. So, yeah, we the conference is, of course, open to anyone. So we try and, yeah, work as transparently as possible, and this is really confounding people sometimes, like every your message, like well, other than what's, maybe a French hat, you know, and then, yes, then some big vision guy, for sure, or maybe the stuff I think of less big vision for sure, but there's not more than that, that's for sure. It's really Right.

Ed: 14:51

So I was talking about free and open source software. I think a lot of people they've heard the words but they don't really know exactly what it means. Can you explain to the audience exactly what is free and open source software?

Jos: 15:05

Yeah, it's a rabbit hole because you can't go very deep. It's the 18th right where a guy had trouble in sprinter, famously, but I think it has changed a bit over time. So there are these two camps. You often hear, like you asked in the 10 quick questions. You asked a few things, but one of them you didn't ask was BSD versus TPL or FreeSuffcer or open source, or even Linus Dorfals versus Vincent Salman, and so these all come to the same point, this question of like, what is the point of open source? And open source you can define as like. I mean, you have an official definition which is really about like, availability of the source, but also what you can do with it. But I think the key question is always like is about freedom and who's freedom?

Jos: 16:03

So free software and free and open source software is about freedom of, often, the user. So it's less about the developer, because if you write software, normally speaking it's yours and you can make it available in whatever terms you want. It's usually protected by copyright. Obviously, all warranties are denounced, right? I mean, if you buy a very expensive piece of software and it does something incredibly stupid, then the license will tell you that sorry, your problem, you know, which is quite amazing give them almost every other product that some warranties but it doesn't. But that's your right as software developer, of course. But the user can be completely locked in by the choice that you make as software developer. You can make webcam software that you can use to check out all the safety of your baby, for example, like a baby monitor, and that stores all those data recordings in a file format that is like I don't know, encrypted and compressed. It cannot be read by anything on anyone unless they pay you money. And like this way you can completely limit people's freedom to do what they want. For example, I don't take out some funny recordings of something that happened and I don't know. Put it on YouTube, or you know what I mean. Like you're in control as software developer, and the point of free software is generally to actually move this access of control over to the user.

Jos: 17:33

And so, most like the basic, most well-known free software licenses, including the ones that NexRoute uses, they guarantee the four freedoms, which are proper nerd style. It starts, of course, counting at zero, so freedom I don't have the order exactly right, but it's more or less like the freedom to run the software wherever and however, you want the freedom to study the software, the freedom to make changes and the freedom to share it with whomever you want. And there's one limitation too, which is usually not immediately mentioned. Right, you have the three, four freedoms. There's one limitation If you distribute it, if you take advantage of the last, the third freedom or the fourth, how you can? You have to share it under the same rules.

Jos: 18:24

So I can give you a piece of free software well, 80 style floppy disk, right or today, give you a download for an app and you can make changes to it and then give it to another friend, but you have to give them the same freedoms that I gave to you, so they should be able to again run it, however, wherever they want. You cannot say look, I don't like you running it for your backpack club. I think people should go in the mountains, I should say in the city. So I don't put a restriction on it and license it. You can't use it for a club. No, you're not allowed to do that. And you're not allowed to take their freedom away to make a piece of it and give those again to other people.

Jos: 19:13

So it's a bit of a viral license, or in the words of the amazing Steve Ballmer cancer. You went they were not a fan of Microsoft in the beginning of this concept to put it out there, because the whole has to be covered under this. So if you make changes, they also have to be covered under this rule. So your mate cannot only then copy on and hand over the software that I gave to you, but also any changes that you made to it fall under the same rules. So of course this does not have anything to do with money.

Ed: 19:47

All of this is about freedom, as in I think a lot of people think freedom means as in kind of free. Free, you know right, and it doesn't it means free, but necessarily.

Jos: 19:58

But, of course, if I give it to you and you give it to 20 other friends and then I decide to go and charge money for it, it's going to get tough for me because, well, all these other friends of yours have the ability to give the way for zero cost. So there is a downward pressure to put it in such a term on the price. So the software itself tends to get cheap or free. However, that's purely the code, and so there is a lot more to code. As it is on GitHub, which, versus an actual running product, there's a big gap. And then this party about code that needs to be compiled. These will be put together. You know you need to make a build. These will be installed and managed, but of course, also everything around it, and it's obvious, because writing the code isn't free.

Jos: 20:48

There's still somebody who got paid to do it, at least in many cases. I mean, we had next clouds, we could have kept this as a, you know, volunteer project, but then we wouldn't have been able to pay like 60 ish full time engineers to work on it, and that's what it takes. I mean, our goal with Next Cloud is to take on Microsoft and Google, right, like we think that it is not good that there are either five or 60 big tech companies that control 95% of all the data that we as a society generate. I mean, we can talk for hours about why it's bad. I would just say I don't think we'll have a democracy much longer if it continues to be that way. I think it's really a huge risk for freedom with a capital F, and so we want to make an impact on that, and I really appreciate and respect all the projects that try to run, really, as, like, volunteer shows, and I mean an organization like MST gets pretty far, but let's be honest, that's not run by volunteers, is it? You know, in the end, if you want to make a real impact, you need money and you can try and do it all donations and like, if your thing is to save baby seals, you know that is not that hard to get money from the general public. If you talk about, like, the freedom to compile code yourself, most people will just stare at you and think, uh-huh, I have other things to do, so it's very hard to run something like next up on donations.

Jos: 22:28

So we, again, we want to make an impact and for that you need a complete competitive product. You need something that can compete with Microsoft 365 and the speed they're developing and it's the huge speed that's going very fast. And, yeah, we want to give people a real alternative, so we need resources for that. So this fully volunteer based it's all free thing Because fundamentally doesn't work for that. Thankfully, free software absolutely doesn't preclude building a business around it, and if you think of Red Hat, that's a free software company, right, what they do is all open source, but it's a billion dollar company, right, multi-billion dollar company. So it's possible and it is a bit harder to do this around open source.

Ed: 23:18

What are the kind of challenges with an open source company compared to, like a closed source company? Lots of challenges, yes, business model and other businesses.

Jos: 23:28

What is it explaining? Honestly, because a lot of people they look at it from a very like technology point of view and they think, well, yeah, but the code, the lines of code the way you know, like as it is there on GitHub, I can just download it. It doesn't cost money. Why would I pay you? So there's difference between here's a bunch of codes and is a product that, as a bigger company, you can actually run and that has everything you need for the success of you as a company. You know, if you're running for 100,000 users and it's down for eight hours while you're googling how to fix a problem, will you still have a job? If you're the one who recommended this to your boss, this isn't going to go well, this doesn't work. And we have what we call an excellent enterprise and it's basically the product that gets you a successful deployment, which comes with our support and documentation expertise around large-scale deployments, again in performance on larger systems. And the hard part. The problem that you asked for is effectively to explain this difference to people. It's often that, like you know the C-level people at the company, they often get it and especially governments often have rules about it. They're not going to run software that's just randomly being pulled from the internet and just pray and hope that it works right. We need something. But there are also companies that don't get it and, yeah, the problem is that, of course, anybody who can download NextLoud and run it on something might think they have the expertise to run it for 100k users, and so there are quite some companies who offer services and support around NextLoud. Some of these then engage with us. They become partners. We sell as our enterprise product, basically, and so if their customers run and do a problem, they will try to fix it first. Right, they do the first-left support. They help you like how to upload a file, kind of problems. But if it gets hard, then you know the person who developed LDAP will be the one who fixes it. You know what I mean. Like, this gets you guaranteed to fix at a fast possible time. Can't say it's all fixed in two hours, but nobody can fix it faster, obviously.

Jos: 25:52

But there are also a lot of especially smaller companies that offer it to other smaller companies. Yeah, and if they get in trouble, I mean, we sometimes talk to these providers and they say well, you know, I'm getting paid. You know, 10 bucks per user per month, and if there's a problem I'll Google it. I'll ask your engineers on GitHub. Right, get free support. And yeah, if that doesn't fix it and the customer gets angry, I give them two months for free. And if they then leave, then so be it. And I mean I get that works for them as a business. But this is bad for the brand Next Cloud, because those customers first of all probably thought that they paid for Next Cloud, but they didn't. They paid for a random person.

Ed: 26:35

So I mean another company kind of setting yes setting the production support can basically give that whole user to have a bad experience. That is the experience they have in their mind about next, and I think that's another thing that they should ignore- it's not just a spotless.

Jos: 26:54

You've seen a test. Well, as I'm trying to say, like you cannot separate the code from the product experience you have. And when you think you can, you just download it and run it. I think it's all fine. But you know your users, who are not technical like you, when don't think in terms of lines of code, they might end up having no pleasant experience because you didn't realize there's more to it. And yeah, this is annoying Our support team.

Jos: 27:22

They get contact requests from people who are like yeah, they paid for Next Cloud support and my problem is not getting fixed. And can you fix it? And like, look, sorry, you didn't pay us. You didn't talk to an authorized reseller, you just paid a random guy or girl on the internet. I mean, what did you expect? Yeah, sure, it was cheaper, but you paid for a whole air. Right, that's kind of the issue. You're paying for something but you're not getting it.

Jos: 27:49

Because in the end, in a way, next Cloud is like Next Cloud Enterprise is. In a way, it's a bit of like an insurance. You know, like if something goes wrong, we will fix it, no matter the cost, and I mean the health insurance that only pays up up to a thousand bucks. That's not really health insurance, right? Because up to a thousand bucks you could cover yourself, but if you get a serious car accident or a serious disease, it's going to cost way more and suddenly you're standing there, you're in the work you should actually have covered and so it's the same with this.

Jos: 28:19

This is definitely a challenge when you're running an open source company. I think and this is something I'm sure other open source companies see too is just explaining the value of the product versus the lines of code. This separation is just a tough, tough conversation. And, of course, for home users, we as a company don't try to monetize that because, like I, cannot help every home user and, honestly, we want to make NextLile good enough and easy enough to use and run that a home user doesn't need us, or a small business, right? If you five or 10 people, you will notice we're not even selling anything to you, right? Our websites, our subscription for an accident, the price starts at 100 users and it's not even that cheap at that level. You need to be bigger before it becomes more affordable.

Jos: 29:08

So, because we designed NextLile to be quite easy to use and to run in a basic scenario for 50, 20, 30 users, it should be fine. If you really want to run it yourself, you should be able to. You can ask for help on the forums and you can ask for help on GitHub if there's a problem and you will not hear anybody complain. If you try to do this for 10,000 users, well, honestly, first of all, that's not very smart and, yeah, it's not really cool to then go into our community or home users and try to get them to help you with your problem. That they would never have. So it's, yeah, trying to balance here between the home users for whom we can start this project. Yeah, right, at the first level I'd be home user too, but for businesses they pay the bills.

Ed: 29:58

Many of our listeners. They run their own science in their own they run on rate science, so they're quite passionate about posting dates on them too. But sometimes they may not, you know, have really talked about actually features in the next day or so, right? So could you briefly explain what NextFile offers, both to the companies and the kind of self-adjustment home, what they're like with the problems, things for the next project?

Jos: 30:24

As a replacement for Dropbox. You can upload your files, you can share them either with other users or via public link, and this this was the core functionality for a long time. But we always had a app infrastructure where people could build apps and already in like 2013, they came calendar app and a contacts app, and so this became really a big deal. A lot of people use that. So when we started NextFiles, we knew there was a lot of demand for these functionalities and, more than that, the industry was moving from just sharing the file to like working on a file. Collaboration became important. So we've been developing since 2016, a huge array of really collaboration platform features. So I think what Microsoft 365 writes you can edit your documents, you can share them, you can have video call and edit them in the video call, and these are all things that NextFiles can do. So it's really a wide array of like it's email, calendar and contacts, it's video call and chat with NextFiles. I think a lot of people, a lot of users.

Ed: 31:37

They kind of think that NextFiles is only file sharing, right, it's all. We have that thing like NextFiles for a lot and we use it internally.

Jos: 31:46

Right, it's a complete platform on which, most ever, businesses can probably do all their like back office work. You know all the spreadsheeting and mailing and internal chat. You can all do it. I mean we do webinars with hundreds of people on NextFilesorg ourselves. We've even done a conference using NextFilesorg.

Ed: 32:07

The NextFilesorg. Should someone want to talk to someone who doesn't have NextFilesorg themselves, or do, they can send them a link and so then that person can join in through their own self hosting.

Jos: 32:20

Exactly, yeah, yeah, so it's the same with NextFiles. You know you can send the public link to a chat room and then people can join the chat room. You know, to get a little pop up to search please give your name that they can join and they can jump there and it's persistentgortz Relation to lead it. You have access, right. So you can just say like, okay, there's guests I don't know who it is kick them out or take away their rights to speak and to turn off the camera and all that. So you have a lot of like controls there, because, of course, first of all, yeah, we do this for privacy control. So there are a lot of like features in NextFiles, like expiration dates and passwords on shares, which we already did like in the early 2010s, while you know these are more recent features by Dropbox and others.

Jos: 33:08

We have, for example, a really cool feature that kind of brings files and talk together that you might know this from banks.

Jos: 33:16

So, like at a bank, they need to send you an email, but they don't want to send it to you because then it goes to your Gmail and Google County did right. So instead they send you a message that says, hey, there's a mail in your inbox. Now you need to go there and you can see there then the emails. So, together with our Outlook integration, actually have a similar feature where the email is actually stored on NextFiles and then an email is sent. Another feature there that we have is we call it video verification, and that's on the integration with talk, where I would send you a link and the link asked for a password, but I did not send you the password. Instead, there's a button that says call. It'll start a video call with me and then I'll give you access. So basically, first face to face, we talk with video call before I just let you in it's a real, I think and then really bringing these things together.

Ed: 34:12

And I was lucky enough to see a video that you are making into NextFile. I'd have seven years today. I believe in talk. There's reactions.

Jos: 34:23

now Can you just oh, yeah, I thought it initially like just the thumbs up to a chat message, but now also, indeed, in a video call, like you can do a thumbs up or a party or a thing, and they can fly over the street. That's really cool. Yeah, these little features, you know, you can do a note yourself. There's menu on each message and you can then turn the message into a task etc. A reminder. On the message you can say hey, you know, 24 hours from now, I want to get a whose notification on my phone reminding me of this chat message. So you don't need to deal with it immediately. You can even do that on a file.

Jos: 35:03

By the way, this reminds me we have files, we have a chat messages and on emails and snoozing email. So that's really cool. And, yeah, you can also forward the message to somebody else or to your own notes. So turn it into a note. Now you have chat to say signal has this do I think, like a note itself where, yeah, totally you win the rule, but you can put all these messages there. So it's really a ton of, yeah, useful stuff.

Ed: 35:32

You can share some kind of insights into use cases for next cloud that may be a little bit different out of the box use cases we all know. Basically companies use this and I believe government choose educational institution out of this. Lots of big companies use it. But what are some of the more obscure use cases you may have across your using that scenario.

Jos: 35:56

So then we first give a big one. I mean it's not super common just because it big. There's a really big Docker provider here with a very bright pinkish color that has several millions of users on a single instance, which I mean they give every customer of them 15 gigabyte of think of free storage space there, so that's kind of an interesting case. Of course, they also are adding more functionality now, since getting more and more interesting what what users get with their phone contract basically. And then what I thought was always a really interesting use case was a couple that was getting married.

Jos: 36:38

They took a next cloud instance on a Raspberry Pi, which is one of the Python Wi-Fi, and they put it at their wedding, you know, like in the room, and they asked all the guests to connect to it. Now, when you would connect, you know there's always this like capture portal that you get, but this capture portal would basically go to a public link where people could drop photos in. So they asked everybody in the audience to upload their photos to this Raspberry Pi so they would get all the wedding pictures, you know, at the end of the wedding there and they told me there was a massive other benefit to it. They did it because they wanted to prevent ugly pictures, you know, getting posted by grandpa on Facebook going out. But because everybody was connected to the Raspberry Pi, they had the internet connection. There were also no notifications going off, you know, no phone calls during the event because nobody had internet. So that was nice.

Ed: 37:35

Yeah, absolutely.

Jos: 37:38

Yeah, that's a really creative use case.

Ed: 37:42

As well. It's just amazing that you know I'm next to have to run on such a variety of different hardware. You know, like a Raspberry Pi up to you know, and really you'd be asked as a service. Yes.

Jos: 37:53

Yeah, but again, that's really important for us. That's also why we are still maintaining SQLite support. You know people often say to us yeah, but SQLite, what is it good for? Well, if you just have a super simple LAN stack, you don't even have an M in LAN, right? You just have a very basic online service that you pay just a few bucks a month for where you can just put in PHP files, a place where WordTresp works, but almost nothing else. Well, next up will work too, and that's really important for us because, again, we do this so that everybody can use it. So, yeah, put in effort to keep it easy to use, even though we're adding all kinds of SQLite files.

Jos: 38:33

These days, that requires, unfortunately, you know, other languages and PHP, right? Certain things you want to do cannot be done on PHP, like example, on push notifications. So, in Excel out, the moment the file changes, this decimal client starts thinking because it can keep a constant connection open with an Excel out and that requires a little team in Go and it requires Redis, because that's how it knows when something changes. And I would count this do that at your PHP, because you can't keep connection open for that long. So, yeah, if you just run the super basic version, then you'll have to wait up to 30 seconds before the sub-sync game and similar, with, like a notification on your phone. You know we support who's notifications, but again, yeah, it can take 30 seconds. If you don't have this little demon running, it'll work because that's important, right. There is always a fallback to just pure next clouds.

Ed: 39:32

So you know, next out runs on a lot of different hardware, but probably the hardest hardware and the vertical is the guy to run on. There's people who are maybe not tech savvy, like if I was to ask my dad to install next clouds, he probably just wouldn't really know what I'm talking about. He may have had dropbox from one of his friends that he can back our photos. So what is next cloud doing to try to overcome the problem of getting people to be able to use next cloud who aren't tech savvy and they don't want to run their own server? Is there anything for them that you're trying to do to encourage them to go in platform?

Jos: 40:10

Yeah, yeah, this is. This is an interesting challenge, and so we've tried a bunch of things. We had, for example, quite some years ago we had the next out box projects where we built a device together with canonical or Western digital that basically you could buy I think it was a hundred bucks, which came with one terabyte drive. Back then it was pretty good deal to a tiny bit of assembly. But just put the simplistic parts together and putting in a chip and then, yeah, you could just go from your browser. You had the running next cloud. Basically that was updating automatically those two snaps and yeah, that was nice, but it's still. You know, you would have to open ports in the firewall. I think your dad thinks that the firewall is a thing that you see in movies with a hacker doing weird stuff. You know what I mean. Like, this is not a solution for still non technical people. I mean, it's not a lot of problems.

Ed: 41:06

I think my time at the firewall was part of five minutes, exactly, yeah, exactly, yeah, yeah.

Jos: 41:13

So I'm going to open that up to be able to run next while I'm on the internet. Yeah, this is a long-term challenge that we've had, and one solution we came up with a number of years ago is we call it simple sign up. So what I call was to make it as easy to sign up to NextLoud as it is to Dropbox and still be decentralized. So how does it work with Dropbox? Well, you must likely will go to the app store and look for Dropbox, download the app, say I don't have an account yet, sign up, give an email address, give a password, accept their outrageous terms of service, which you don't read anyway, and then you load in and you have like two gigs of research. So what we did is we have an experience where, in search for NextLoud on the app store, you download this, all the app and open it. You say I don't have an account yet, sign me up. You give a username, you give a password, you sign much less outrageous terms of service and you get between two and five gigabyte free storage.

Jos: 42:19

So how do we do that? Because we don't host right. We don't want to do hosting. We make software stick to what you're good at, what we have done at the first step. When you entered your email address, we also looked at your location, based on the address, and we picked a provider for you as close to you physically as possible. That's part of the program, and so these providers then give you free accounts, unlimited in time, but with between two and five gigabyte free storage, and I mean usually campaign them to get more storage, which is how they make the money. So, accomplished, we have the same user experience as Drobot.

Ed: 42:58

So that's really cool. So in very no set up involved at all. Exactly, you're just looking at.

Jos: 43:06

Yeah, you can scroll down from where you entered your username password. That's like picking up. So it's not reducing a BPA, then Right, yeah, you can also do this via the website. So you can just go to our website and there's a sign up and you can just there also sign up and then you get the information, the URL.

Ed: 43:23

I'll make sure to put a link and then show it out to the others.

Jos: 43:26

Yeah, we had out this. This is really nice and we're proud of having been able to make that happen.

Ed: 43:31

So the other question, just looping back to the open source list, and it's about, obviously, community. That is a big role in it. If someone in our audience actually wants to get involved, how would they actually go?

Jos: 43:45

about doing that. So it depends a bit what you want to do, right. But if you want to learn coding or design, then we have a page on our website next up called contribute, where you can basically find guides, tutorials and other stuff. You have to build an app, which is now possible in any language you want, exactly yeah, so you can build the next app, still the traditional way in PHP. But since the conference earlier this year, we announced that you can now build apps in Ruby, go, python, whatever, haskell if you really want for Cocoal. So we support pretty much any language.

Jos: 44:29

It has a restricted set of abilities, but we're working, of course, to expand that. And then for the Hub7 release, we also made it super easy to install these apps. So if you say, hey, I definitely go, I want to build an extra app, well, go to nextcom. You can find tons of tutorials and information. Or if you say I have an app I wrote, like I don't know, a notes application, and they want to make it possible to store these notes in next loud, for this we have the open collaboration services, ocs API, which makes it super easy to connect to an Excel server, you know, retrieve login data and get those files then stored there so to connect to an Excel service. We also have tons of documentation on that.

Ed: 45:16

That sounds awesome. So there's a lot of talk nowadays about AI and NetSplot as a brace AI into it's system. Can you explain a little bit about what makes NetSplot different with its use of AI compared to other solutions out there?

Jos: 45:37

Well, in the nutshell, what makes a difference for everyone else? It's, you know, under your control. So interesting fact is, we've been doing AI like an AI is, of course, like a bus word. So in our marketing I usually group all stuff under AI, simply because for people technical side of the matter, so if it is smart I'll call it AI. You know, like I don't know, recommending people to share with, but that's of course not using a neural network, that's just using statistics of who you share with a lot. But we've actually been using like a BSP library that lets you train and run a neural network for at least two years, maybe even longer. So we use this to log logins of users. So for each user, we record when they log in and where they log in from and I think the client to or something it's like two or three factors, and we train a neural network on that and when you do something weird we'll give a warning or can even, you know, warn the administrator. So what this does is basically you know and obviously fucking interrupts.

Ed: 46:52

This is all that we use as our server. It's not exactly Exactly.

Jos: 46:56

Exactly, yeah, so this runs on your own desktop server and it's a very simple small thing that we've had for years. We call it suspicious login detection because that's what it does. Is basically trains this network on, like where you typically log in. And let's say you work night to five every day from the office at Bristol and suddenly there's a login at 3am from China. That's probably not you, so you would get an application for suspicious login detection and as admin you can even say let's block that and that's a neural network. That's pretty fancy AI, but it's been there. It's been there for years. It works in the background, so probably barely real lines that that is technically AI.

Jos: 47:40

Another thing is we've had like a priority inbox for quite a while where, again on your server, a neural network is trained on your emails. We've recently improved this. We're also training on the subject line so that it's getting better at basically predicting whether it is an important email for you or not. And then, yeah, you have this inbox that has hopefully the most important emails on top there. Yeah, completely on-prem AI. And then last year we also introduced face and object recognition in the Nextapp Photos app. So you have your photos and then it will go over it and group the faces and name objects and you can then say, oh, that face, that's my dad. So you just give it the name and then you will see. And if you search for your dad in the search you will see all the pictures of your dad. So this feature was also already there for a year and again all running locally Because this can't have the photographs?

Ed: 48:35

Does it have various things or other than it knows what it is? Or do you have to say it, dog, like? Put it into me, it's pre-trained. It's pre-trained, that's right. I hope you can have to say that as a dog.

Jos: 48:47

No, no, because you would have to tag thousands of images of dogs before it would get it. That would be a bit too much. No, no, we. So it has the parts.

Ed: 48:54

So they're only the peak and the target people Exactly yes, it's already trained on recognizing faces and grouping them.

Jos: 49:02

So it'll group the faces and you get a folder with pictures of, hopefully, the same person. It's not perfect of course, it might make mistakes, but it's fairly okay at it. And then you can change the name and then groups and dogs and airplanes and trees it'll recognize and tag, basically, so you don't find them.

Ed: 49:21

So I'm assuming, with AR features in NextCloud, this is where you're going to need maybe a more powerful system to get good results and maybe win on Raspberry Pi Zero.

Jos: 49:36

Yeah, so the suspicious login detection and the priority inputs are pretty light. These were running already for years and they were fine. You don't need anything special. When we introduced the images and object recognition, this app needs like a neural network which is like a firewall of about a gigabyte big. So we already wanted to do a first problem. We now need to download a gigabyte data and a 60-second time mode on a PHP process is going to be paid. So this was already where we had some more trouble to make it work. And then, of course, came well end of last year where Chachi came and we said to each other okay, on one hand, we want to offer the abilities that this brings and try and allow people to benefit from it, but it has to fit on DNA, it has to be transparent and on-prem as far as possible, but then you do want to give people a choice, right, because running a large language model on your server if you have Raspberry Pi, it's just not going to happen. It's just not going to work, period. So we spent quite a bit of time thinking about this and we came up with a rating system for AI. So we are supporting these new capabilities like language generation voice recognition. What else do we have? Add transcription, of course. Right, you can do a talk call, you can record it and then at the end the recording can run through a voice recognition speech to text. All these features we have, and in most cases we have them in multiple variants, with usually one of them being like open AI, for example, right, which offers that as a service, which is, yeah, I mean, it's fine for some stuff. The further stuff you don't want to send your data to open AI. And so when we put these choices in the app store and we put in a what we call the ethical AI rating, basically each of the services we rate on three points, namely is the code that runs it and builds it and trains it. Is an open source, so that you can actually, you know, retrain and build and check and look if there's no nasty stuff in there. Is the model available to actually run locally Because if you can't run it locally, then even source code helps you much right, if it runs in the data center. And, last but not least, is a training data set open so that you can actually, with the open source code, you can retrain the data model, make it maybe more efficient, make it smaller to run on your system, remove bias, right. There's tons of like crazy stuff in these models. That's not always cool With biases and other issues.

Jos: 52:40

If the training data is open, you could do something about it. So we date these factors and we say okay, if it complies with all three, its open source, can run wherever you want and the training data is open and transparent. That is green. If it doesn't have all three but only two of them. So let's say you can run it locally and the code is open source, but the training data is proprietary. Let's say the training data is the database of Spotify Is it the song recognition AI? Just to match your salt in some way random, then it's orange. If it only has one of these three, then it's yellow and if there's none of them, it's red. So JetGPD is obviously red, but GPT for all, which is a local runnable AI model that is completely open source. Training data et cetera, will be green.

Jos: 53:34

And then you get a choice what do you run, what are you installing your server and then all your server. You choose what model is used for what. So you can say look, my images are generated by an external service, because that's fine. But I want call transcripts to be done locally. So voice to text needs to be done locally, because I have my calls. I don't want that stuff to leak. And then text generation well, we do nothing special and JetGPD is great, so we'll use. You know what I mean. You have a choice.

Ed: 54:05

You just have a hybrid, you choose exactly what you want and nothing's installed by default unless you want it. I mean no in the raters system. What's your name? That's?

Jos: 54:17

awesome, yes, yeah.

Ed: 54:21

How do you go about external AIs? Obviously it is cost to use them. If you had like a bunch of users, is there any way of making set A cards Right? Yeah, it costs like lots of money.

Jos: 54:35

Yeah, you can put limitations on the use, so you can have a per user limit and you can put that in. So, so many tokens for tax and so many images for the images yeah. But that's even useful, I think, locally, because if you're a university with like 50,000 students and you say, hey, that model, that's loud, no hands that can generate text for us, turn that off because we got a couple of servers with a few GPUs and them, so great. And you know how can 50,000 students think, oh, that's great, I'd love to have my next, you know work that I need to deliver tomorrow be edited and summarized or even just written by our GPT for all model, then you know your server will be actually busy.

Ed: 55:23

Yeah, I can say that I'm not Mr Posting, it's almost exactly. Yeah, it's pretty crazy yeah.

Jos: 55:28

I mean this stuff is really heavy. I mean you have to choice of models, like a Lamma model we support, and then you can choose like a smaller, the medium or the big one. At home I've played with the medium one and that's working pretty okay. I mean the general idea for tax. It can make headlines and summarise the text. If you ask it, you know, to write you a blog post. It's not great, but like that's okay with me.

Ed: 55:54

If you feel like a home user, what power of hardware would be needed for adding a lot of the best AI models? Now, what do you consider the minimum?

Jos: 56:05

Yeah, I mean at the Raspberry Pi, it probably won't run because one of the problems is it needs a lot of RAM. These models I have at home a Haswell system, so that's not the newest quad core i5 and it has 16 gigs of RAM and I haven't bumped into hardware problems. It does take a minute or two sometimes to generate like that apply or an invention or stuff. So it's not fast, but it's fine for me and our UI is pretty optimised for, like you know. So when you do a request and you say, okay, draw me a picture of two dogs and I don't know whatever, then yeah, you can just basically tell it to background itself. You just get picked on your phone or on your desktop. You know like, hey, if it's generated, you click to see the results. So the user experience, even when it's slow, it's fairly okay. I mean, when you sit there waiting for it, then yeah, that's not great, but if it takes a few minutes anyway, you can just say, okay, well, I'll move on, I do my other stuff and I'll see notification coming when it's done. So I would say, as long as you have enough RAM, you can run anything, and the limit is going to be your patience, because of course it's not fast, but most of them do support hardware acceleration so you can have your CPU Exactly, yeah, yeah.

Jos: 57:29

So if you've got a 4090 with a ton of RAM, then you can still not run the biggest models, but you can get pretty far, obviously. And then, yeah, you can totally run this. And so the all-in-one container, which is what I'm using myself since a few months, is really amazing and it supports all of this. So, like it can spin up a container for local AI and in there then you can download a bunch of models. You just need to put some magic, a JSON file, and then in the JSON you put like, the models you want and then when you fire up the container it downloads them and then they're available in UI to choose from. It's really neat. And then you can just enable hardware acceleration. Then that's just a parameter when you start up the container.

Ed: 58:15

And I know I'm doing it when I get back to Bristol.

Jos: 58:17

Yeah, look, it took me some days to get it working, but I'm not very technical, although I must have been to have a cheat line to actually do it, with the people who work on it, obviously. But yeah, I got it working and I can tell you it's freaking amazing, it's so cool. So, yeah, I mean, all this stuff, big Tech is trillions at you know, yeah, you can have it, I mean.

Jos: 58:44

Yeah, I just think they think I have anything in chat right and the big microsoft, the Google I have a review on that part but we're not limited to Big Tech. You know we can do this stuff and I think it is beautiful to see a ton of models being developed. It is, of course, tough because making a model costs, you know, several hundred k's until several millions just to run the training data through the model, to get all the weights and to basically have a trained AI. But there are a ton of people who basically done this already. So you have these based models and you can build it on them, and this happens a lot in the open source worlds. There are a ton of interesting, cool open source models that can do all kinds of really good stuff, that are freely available and can be run on your server.

Jos: 59:36

And what we're working on if I can pitch a really cool feature that brand new one have seven obvious question wouldn't that be great if your AI knew you right? So what we've built is we call it context chat. So we have the next other systems and this is like one of the premier ways of which you could interact with, especially the large language models. So an extra little top right, if you have it enabled, you have the little button that calls up the assistant. The assistant that gives you a dialogue where you can ask it something so you can integrate it in the next text. So, for example, you could just select a piece of text and call up the assistant and say summarize it for me or, you know, generate a headline and it will give you the headline. And so, with context, chat, the questions that you ask can actually take into account the data you have. So we have a service that indexes your data. All your documents, your files are not there yet, but we want also emails, chat messages, everything else. And then you can basically say hey, what are the tasks that my boss sent to me by email and chat in the last week and it'll just take that out of all your emails and chat, make you summary, say, ok, your boss asked you to do one, two, three, four, five, these things you know, and super cool, that's pretty good.

Jos: 1:00:56

Yeah, so we have a tech preview of that available by the time this podcast comes out. I mean, you've got to be pretty techy and tenacious to get up and running. I can already tell you that because, like, by the time I got my AI stuff up and running, it was already beyond tech preview, like it's already a little bit polished. I say a little bit. There's a lot more to do there. Well, this is really rough, but it's possible. I've seen it that way. I've seen it work. It really does work and I think it's really awesome. And this is really the next step. Yeah, then in your AI, you can just help you with your work, because it knows you and your work. I think that's really amazing.

Ed: 1:01:35

Wow, yeah, I think that's something that a lot of people has had no idea about, like some next cloud or a secret site.

Jos: 1:01:41

Yeah, On your own server right Now data leaking. It just stays there.

Ed: 1:01:47

And things you want to allow software to use to help out.

Jos: 1:01:50

Right, of course, because there's also work with an external AI. So you can actually have this database and it's called a vector database with all your data that needs to be generated by next slide. You can have that local but then use like chat, gpt or a local service. There are companies that are now starting to offer large language models as a service. So you can just find a local company actually you do trust. Just gives you the choice Exactly, yeah, and then you can connect that and then that AI from that company that's hosted maybe in your city, you know, can answer your questions. You know you basically pay for the compute and you get your questions answered based on your data. This is possible.

Ed: 1:02:33

Yeah, well, anyway, switching gears a little bit, I've got question. Next bout had files and anti-competitive lawsuit. To get to Microsoft.

Jos: 1:02:44

Right, yeah, it's a complaint, so it's not directly a lawsuit. Right as to them, because I mean they have a few more lawyers that we could, that we have engineers, so you know there wouldn't be a fun fight. No, we found a complaint because the EU has anti-trust laws. Yeah, if you look at the integration of, let's say, one drive in Windows, yeah, good luck not signing up, basically as a normal person who were Dropbox or Next Cloud. That's unfair and Microsoft is using their weight in the desktop market to basically get into the file sink and share market. That's almost the definition. If you have this market share as they do, it's almost the definition of anti-competitive behavior.

Ed: 1:03:29

I think especially as well when there's an event, you pretty much ask the sign up to the likes of kind of to that, unless you want to.

Jos: 1:03:36

Right, and if you have Microsoft accounts, you have a one drive account. That's how it works. They explicitly say that to you, with documentation everywhere else. So basically, not having one drive is even a possibility if you use Windows and, let's be honest, windows is perfectly usable without one drive. So that is explicitly done to squeeze hours out of the market, including us. So the EU ought to do something about it.

Ed: 1:04:00

And what are the main outcomes? That that's out because they're going to achieve a bit of end.

Jos: 1:04:06

Left of playing the foods. Yeah, I mean. No, I don't like me. Microsoft will stop doing one drive with a fact that they're an equal shield, and whether that means they stop bundling any Windows and forcing it down the throat or they offer you the choice of multiple options. Either way it works for me, but it needs to be fair. It just isn't now.

Ed: 1:04:30

You know playing on devils. I have to know what would you say to people who say Microsoft is simply offering convenience for the user.

Jos: 1:04:38

Oh, they are, but they should do that in the world. They're doing this, of course, not because it's convenient, because there are tons of features that they don't offer it, though they will be convenient. I mean, come on, there are so many anti features, right? I'm still one of the things I always doubted with Zoom. If you join the Zoom call in the browser, it'll tell you to download and open the app and only once you've tried that two or three times and it failed, it'll suddenly show you the ability to join by browser. Yeah, why that's not convenient? That's just because they're trying to get their spyware installed on your system so they can you know? Yeah, it is, but Microsoft does the same stuff, right, and Dropbox and others too.

Jos: 1:05:24

Like with Google, if you share a file, you can only on this file, say I make it available publicly, that anybody with the link can use it, or I only share it with a couple of people. Specifically, with NextFiles, you can make three public share links. One of them has write access, one of them is read, only one of them. You can only edit the file in the browser, but you cannot download it. You know what I mean and send these three links to different people. It's not possible at Google. Why, well, they want you to create an account, they want everybody to create an account and then you can have granular permissions like that. Is it a technical possibility? Well, hello, we're a tiny company here in Germany, we can do it, so they can do it, but they just don't want to.

Jos: 1:06:12

So the argument Microsoft is just make it easier. It's like well, we all know that's not the primary reason they're getting into this market, and if they wanted to make it easy, they could have give you a choice. They could say hey, install any app and it'll look nicely into our ecosystem. So, for example, if you're using Microsoft Office on your iPad, you want to save the file. They can save directly to NextFiles, to Dropbox, to Google Drive. No, you can't. You can only save directly to OneDrive. It's not possible to save to any others. You for that have to export the file, but in the meantime it'll still put the versions on OneDrive, even if you didn't ask for that, and you can't replace that. It's not possible. So again, convenience, no way.

Ed: 1:07:03

I was actually using Word for the first time in many years the other day and I just sold the top. There's a little top over where you can have a photo to save. I've got no ability to do that because my back's too crowded a bit. I was going to make sure. On top of that, he goes you must have a cloud version of the file first. And I was thinking what? Yes, why? Why do I need a cloud version? It's all too safe to my hard drive.

Jos: 1:07:30

No, no it's going to the clouds and they want it. And again, do you have an option there to use NextFiles as the backend for this? I'd be surprised. And so this is not convenient. This is anti-competitive behavior. They'll definitely defend it as convenience and they will do the same thing as they did with Internet Explorer. You know they were going to get forced to take Internet Explorer out of Windows, so they then build it in so that it was used as a desktop to render the desktop background. Not that anybody wants a web page as a background. You want a picture and I put on it. But they've built in Internet Explorer so deep that when they were forced to take it out, they said, oh no, we can't. Well, that's called strategic incapability. You know so, and they're going to do the same here. I bet that, like this whole out of safe feature is going to be used as a.

Ed: 1:08:22

I didn't do that. I didn't do that, so that's why you have that kind of active web desktop and then the whole Windows.

Jos: 1:08:27

Yeah, just anti-competitive stuff. Yeah, we're asking for that. Right, we'll need some browsers, and if the user mattered, they would be doing. You know, Windows 8 was all about getting into the template and touch markets. Did they do that to make the user's life easier? Heck, no, they did it because they wanted to go after Apple. And the users are just. I mean, you're there to give them money, but otherwise you're not there. They are not there for you. I think they make that crystal clear at every corner.

Ed: 1:08:59

I have some of the segments not the companies that could be used as an obstacle to the wallet. Yeah, exactly.

Jos: 1:09:05

And that sounds like.

Ed: 1:09:06

Microsoft. Yeah, and it's nice to get company that puts you first.

Jos: 1:09:12

Yes, yeah, well, yeah, but again, that's our private motivation, right? We do this because there needs to be a genuine alternative to these big companies. For that you need to have a good product, because otherwise nobody wants to use it anyway. And the next cloud is I don't want to say it's perfect. Even product has flaws, and certainly next cloud does. But we have customers who complain that they're in sysadmins and they complain that their users like next cloud so much that they're pushing to roll it out to more departments even though they're not ready for that yet. It's just a good kind of complaint to get, in my opinion, right, so yeah.

Ed: 1:09:54

Switching gears. I've got a couple of user questions that I've posted on the forums or, if I can pose them to you, yes, should. I've got a question here from user Johnny O or Johnny Zero, I'm not sure. He says why is next cloud slower across platforms, even on high-end software, and are the plans to improve this? Are the plans to switch from PHP to another programming language for next cloud? I think I was spoken to that a bit and will there be a slimmed-out version of next cloud to just use the specific parts of it?

Jos: 1:10:27

Yeah, I can answer that. So, performance-wise, this is as I said earlier. Next, cloud is complicated and if you've set it up well and are all in one makes it a lot easier. Honestly is very fast, but yeah, that can be tricky. Look, I don't like to blame the user because I'm sure they know what they're doing. Honestly, my private next cell is also pretty slow and I haven't figured out how to fix that either.

Ed: 1:11:02

Yes as well. It's not just the code at Next Cloud being slow. If you really pour into that connection around the next slide etc. My mum's house, I think she's got two megs down and half a meg out. I'm not going to get a big connection trying to send photos of what happens.

Jos: 1:11:18

Yeah, they could play a whole bit, but let's assume because they sent, you know, even on heavy hardware, so let's assume it's also in data center and fast. There is really a lot of like you know, the right readdisk configuration, the right, you know, patch your engine next configuration, but also our hyper folder backends, you know, like you have the type folder backend for files. There's the push notification demon that I mentioned earlier, the thing and code. These parts all make a big difference and it all has to be set up the right way. And honestly, I can't speak the wired slow in a specific case, but I can guarantee you it can be just very fast, just as fast as like the Microsoft and Google or other products. And so it's not the codes, certainly not the PHP. That's the issue, because PHP has its browser cons. But performance is not a problem of PHP, because it's pretty fast and it's what's the biggest benefit. Of course it's really scalable, right? Because if you want to do four times as much, then you just quadruple the number of PHP servers and you get almost 100% scalability. That's what PHP is good at. The problem for scalability is always the database and the file system, and those are also usually the performance bottlenecks. Of course I'm sure there's room to improve things in Nexar as well. We work a lot of performance. Always, as I said, we have like components in Go. There's a Rust component to Nexflot, other pieces, so to get it as fast as possible you should turn all of that on and install it and configure it, etc. Or use the only one, and that will make a difference. But while there are certain things that PHP can't do, if we would basically stop developing Nexflot for two years to rewrite the whole thing from scratch in Go and then another three years of like stabilizing it all again and fix all the bugs that we've already fixed in the PHP cards, yes, we would be five years behind on everybody else in the industry, but now it would be really faster. It might even be slower because it's a lot of these things are in architecture and the language is really the least important thing, the least important part of it.

Jos: 1:13:27

It's the same conversation that you can have about ARM versus X86. It's not the instruction set that makes a difference. Yes, arm systems are more energy efficient, but if Apple had a license to the X86 cards, believe me they would be the same laptops and the same leaders and performance and they would be equally good. Because they put Jim Keller on it and a bunch of other brilliant people and they made just an unlimited budget chip that is massively huge and runs at low speed and is super efficient, has nothing to do with the fact that they're using an architecture set Well, nothing, but it's like 1%. It's the same with PHP. Maybe it'll be half a percent or 1% faster, but an architectural choice of like you know how you store things in the database or you know how you access the files, it's got to have a 10x, 100x impact compared to the language itself, so it just wouldn't make the impact that you might think it does. It's really a micro benchmark thing where it makes a difference, but PHP is perfectly capable of being incredibly fast.

Jos: 1:14:39

The last point was about making a slim down version of it. Well, you can do that, because next slide is really it consists of apps. I mean, even sharing, which is a pretty core functionality, is an app, but there's a ton you can turn off. If you say, look, I don't use versions, well, turn it off. If you say I don't use group folders, I don't use guest accounts, I don't use external storage, turn the apps off and they will not be there. But the thing is it's not going to make a lot of performance difference because we've been optimizing the bananas on this for well more than a decade, and so it'll make some difference for some areas. Now, obviously, loading the page and then the versions takes longer than loading the page Period. So, yeah, turn off previews, turn off, you know, icons. Yeah, if you make your phone black and white, it'll also use a little bit less power. But I mean, you know, is that really the issue? But it'll help. Sure, yeah, you can do it.

Ed: 1:15:46

So another question from that whole bag for anyone how does the team prioritize? And husbands the test detect, fixes. And he wonders has the team seen a GitHub issue with singing files larger than 100 gigabytes? I believe that's in to do with next out and file tracking. Yeah, with cloud. With cloud I mean cloud and chunk out where chunking?

Jos: 1:16:17

I know that issue because I don't do that discussion myself. So is this tricky? Because, as I said before, we're ambitious, right, we want to take all Microsoft. So, pure, rationally speaking, what you then want is you want to get as close as possible in those of features and usability and accessibility and all these things that you need, you know, also for companies, right. So both for home users, because that's the ones we do this for, but also for companies, because they pay the bill. So you need to balance that and you need to balance the deep getting to these features without, of course, making it unstable for the majority of users.

Jos: 1:16:59

But with bugs, the issue is that there are multiple kinds of bugs in the old range, from like affecting I don't know like a thousand next out servers, which sounds like a lot, but there are several hundred thousands out there, so that's like under 1% still. And there are bugs that affect everyone, and I mean it's obvious that we tackle the bugs from the top of the list, right. We tackle the ones that affect everyone and that have a big user impact. That's the other part, right? I mean there can be a bug like, for example, right, and we don't like to point out bugs, but it irritates me too, so I'm just going to do it. There is this bug in the comment section where you can add mentioned a colleague and it does that right, if you type ads fris, and then it'll add the complete Christine and it'll show you users and you click one and then you post the comment and then it just says there in pure text, ad Christine, and it doesn't render this nice thing that you can click on and then contact her, etc. That's a bug and it affects every single user of next out for quite a while already. That doesn't affect you. It doesn't make you work harder, it's annoying, it's a visual bug. We'd like to get it fixed. But it affects not your day-to-day work. You can still work.

Jos: 1:18:16

And similarly you have bugs that are completely crippling and make the system not work for the three users that use like an external NFS storage together with a Oracle database and have user accounts with space in it. You know there will be a bug like that that affects, like I don't know, three servers out there. Now one of those servers is from a customer that pays us money. It'll get fixed very fast, but if not, I hope you understand, this drops to the bottom of the list. It's just we can't fix everything, and if we only would fix bugs first of all, there will still then be another library that is updated and break. Something like zero bugs will never be there and we will simply lose speed, get further behind the Microsoft, which means less companies will be willing and able to pay for next time because it doesn't work for them anymore. It doesn't do what they need, which means we can't hire as many developers that you get in the downward spiral and the end result is just worse off. There. It might be a little bit more stable for like the six months, but once the effect on less customers starts to kick in, you know what I mean. Yeah, so it's a very fine tuned balance to get the maximum development speed that still allows us to fix the box to the degree that it does the impact most users experience, but still allows us to sign up as many customers as possible to hire more developers to increase the development speed.

Jos: 1:19:44

Because, again, we are trying to keep up with Microsoft here and if you look at the resources they put in Microsoft, these we're growing by 50 to 70% per year and you can say, wow, without fancy capital. Fantastic, that's great. Why don't you slow it down a little to fix more bugs for me? I get that. I mean again, personally I'm annoyed by some bugs too, even on my private server. But Microsoft teams grew by like 800% last year. So we are losing ground at a crazy speed by growing just 50%. And if you look and here I know it's a big picture but if you do a step back and you say where are we at 10 years? We want to be in a space of 10 years where as a company or as a government or as a consumer, you actually have choice, and I don't mean the choice between Microsoft, google or Apple, but I mean the choice between Microsoft, google, apple and Xlash or something else is obviously doesn't have to be us, but I didn't need Xlash. So if we are 0.01% of the size of them by then we will not be a viable alternative. There's just no way commercially to make that happen.

Jos: 1:20:59

Think of phone operating systems. I know you're running a pretty obscure operating system, but what you're running is not a real alternative for the average person. There are two operating systems on the phone that are usable for a normal person. Why is that? Because in the market there just wasn't space for a third, and I mean why Microsoft tried to be the third and they failed. So you can't tell me that this is nothing other than just purely how hard it is to peep in that market.

Jos: 1:21:30

So if, 10 years from now, there is only space for five players in the file storage and collaboration markets, we need to be the number five. Because we're the number six, we will not be a viable alternative, an actual choice, because by then there will be huge vendor lock-in effects, just like what we have in the phone world with the App Store. I'm sure that'll be equivalent things and you actually will not have a choice for next round and we want to be there. So we need to maximize speed and so we need to find this really fine balance between features, bugs, customers, home users and I can tell you we're doing it wrong all the time, because this is a nice edge and obviously we're not exactly in the middle there and you will be able to point me to tons of bugs that maybe on one side or the other. Maybe we're doing it wrong, I know, but I hope it explains our priorities and what we're trying to do here and, again, whether it works, maybe, maybe not, is all an easy balance for sure.

Ed: 1:22:36

Yeah, I hope it's perfect. I've got to get into a big picture and things that are not right and people are before. It can't just.

Jos: 1:22:45

Yes, exactly.

Jos: 1:22:46

And so, again, it's not that we don't care, it's just we're trying to balance.

Jos: 1:22:51

And, again, if I have the choice to help this one customer, because if we fix that or add that feature, you know, like the work on accessibility at the moment, fixing tons of accessibility bugs for people who use screen readers, so more than 350 of these issues have been fixed over the last six months. That's a lot of development resources and tons of testing and finding these and fixing, and we're not yet at the point we can get the certification. I think I mean, you might be this, they're working towards that, the ones we got that the doors will open for a ton of government contracts and that will obviously pay this back. I know, 100 times over, maybe over the next years. I don't know the numbers here, of course, but you can make a calculation there and say, well, yeah, we didn't fix this bug, we didn't fix that bug, but we are investing in this and this will allow us to fix so much more and develop so many more features. But of course, it'll constantly be this, you know.

Ed: 1:23:54

So, yeah, but just one step at a time and taking the right steps.

Jos: 1:23:57

and yes, and again, this is really hard to find this balance. So we're always very open for feedback. But you know, just saying, look, it annoys me, can we do? But we have to look at the big picture here, because I mean, actually it wouldn't be here without the big picture, right, and the company wouldn't work without the big picture. So, yeah, sorry, but like, just the fact that it annoys you cannot be a reason for us to not to not do it or to do it. You know just how it is.

Ed: 1:24:32

I just the tall guy's question about cloudflare. I believe that can be fixed in the actual client by a major chance. Yeah, yeah.

Jos: 1:24:44

So we found a solution for that article from the trend that if you set some limit in the conflict file, then that stops it and look, I hope that's the that's the client people at some point have time to basically allow the supply to figure this limit out by itself and then put it in by itself.

Ed: 1:25:03

Right.

Jos: 1:25:03

Because that should technically be possible. I think somebody even made like a promise of writing that code and then that's just review and prove it. But yeah, that hasn't happened. And I know that the Desupply team has a metric ton of stuff on their plates. I'm married to one of the Desupply developers and I hear the cries. It pains sometimes. Really, they would have fixed it if they could realistically do it. Believe me, they really want to, but they just haven't been able to get to that.

Ed: 1:25:42

Well, josh, I can't thank you enough for the time you've given us today and the insights you've given into the world of Next Found. I want to give also a big thanks to the whole of the Next Found team for passing on. No, you won't the team, so I can say you know very appreciative of the work we're doing, championing privacy, open source and taking control of our data. We need more companies like you and I wish you all a successful upward and I hope in Chandy's time we're not on the side, but I hope you're. Number three tool.

Jos: 1:26:15

Yeah, that would be lovely.

Jos: 1:26:18

Yeah, that would simply mean a lot more people have control of their own data, and that's what we need.

Jos: 1:26:25

Josh, just hope that we then still live under, you know, a not a supreme leader, but just a decent society where we still have choice. Indeed, thanks for coming and really appreciate it, and, I hope also thanks to the community for using Next Found, for advocating for it, because really there's makes a difference, like getting Next Found in companies, explaining to people why it is important and you know what's the benefit of being in control of your data and caring about privacy. We're just making this AI run locally and just prove that it can be done. This all really makes a difference. And, yeah, we wouldn't be where we are without the community, like not just developers, but also people like your community, like the unright community and the listeners who you know fiddle and fix things and report banks and talk about it already to tell other people that this has an impact that made it possible for us as a business also to work and to do what we do. So thank you too.

Ed: 1:27:28

Thank you very much to your listeners for tuning in today. Keep learning how to take control of your digital lives. This is S and you're signing off Bye, bye.