NuNet Makes Decentralised Compute Easier to Use With Its New Appliance

Okay guys, we are looking into NuNet in this video episode and the team have got some pretty cool updates and one of them is around the brand new application that’s been released and this will allow you guys to join in on their decentralized compute and join that network and provide compute power to their network and other people can start using it. This is one of the big things that NuNet has been trying to do for a while and it is now reality and you guys can jump into it. I have Jennifer joining me on this episode to talk through the application, what it means, how you can join it and all these other questions I have around it. So Jennifer, welcome to the podcast.

Thanks for having me. Glad to be back and catch an opportunity. It’s been a long time. It has been a little while and I think we should probably start with that quick elevator pitch of what NuNet is and what you guys are trying to do just for those people that haven’t been keeping up to date or some new people that are just entering into the ecosystem at the moment.

Sure. NuNet is a decentralized computing and orchestration platform. So essentially what that means is from the decentralized compute perspective is that anyone can contribute compute into the network, whether you’re a MacBook user, Windows user, if you have some Raspberry Pis, which I’m currently setting up here, if you’re a data centre, even if you’re another DePIN project that want to push into the network or to allow your compute into network, you can join into NuNet. The other thing that NuNet is, is an orchestration layer.

And basically what this means is that it finds the right compute for the right task. So the way I kind of like to put it is the composer of an orchestra basically tells the instruments when to come in and to come out at specific times or, you know, when and out. And that’s essentially what the orchestration layer does. So this is kind of what makes NuNet very different from other decentralized computing projects is that we’ve actually built this fully decentralized orchestration layer on top of it, which is something really, really cool.

Very cool indeed. Now I know there’s a lot of use cases for this. Could you just explain to the users who could use this platform and who could provide? I know I’ve got this giant ass computer under my desk here and I could potentially join this network.

Is that correct? Correct. Yes. So from a compute provider side, we have very low barriers to entry.

So your MacBook can join. Like I said, edge devices, like I have this Raspberry Pi here that I will push into the network. If you have kind of small GPU mining rigs, whatever it may be, data centres, you can also, we’re working on if you have like a account with Hostinger or any of those that you can actually contribute your spare compute into the network that way as well. And then from a use case perspective on who can use it, it’s very, very, very vast.

And essentially any kind of compute job can be run on this. It’s not just singled into one category or anything like that. So anything can be run on it. So right now, actually what we’re doing is we are running n8n, the NuNet network, and all of our support channels and automation is done actually through n8n via the appliance.

Other use cases that we have that I’m actually was working on today is deploying open flow on NuNet, and then actually using Qwen, the large language model, so that we don’t actually have to tap into Anthropic or OpenAI API or anything like that. So you can run these private AIs on NuNet. That’s more from like a personal standpoint of maybe somebody that would want to use it. As we know, open flow is everywhere.

A lot of people use n8n, anything like that. And then from the other side of things, the more business use cases. So right now we’re actually working with a company that we are connecting edge devices into a single network and allowing for private AI agents to run on them and energy agents. So this is looking at kind of edge AI, where we’re actually picking up on like small little IoT devices, connecting them into one network so that everything can communicate together.

So it’s quite a vast array of things that you can do. Of course, we have our Cardano SPO use case as well, which we’ve had the relay node run on and also working on the block nodes as well. So basically anything that you can dockerize, you can essentially run it on NuNet. That’s like you say, it’s a pretty vast array of different use cases.

Yeah, definitely for sure. And I think that OpenClaw, the AI agents, or the personalized AI agents is a really big one at the moment. We’ve had a meetup here recently on the Gold Coast where a thousand people rocked up to an OpenClaw meetup to learn and understand how to set things up. So having this type of option here as another way to utilize unused CPU power for your OpenClaw agents, I think is brilliant.

And I think a lot of people might start using it just for that. Yeah, yeah, for sure. So I’ve actually moved off of, you know, I had some agents running on cloud services and now that NuNet has their appliance release, I’m actually running, well, OpenClaw through NuNet, as well as I’ve been tinkering with Claude Code as well, having cron jobs on that. Just a piece of Python scripts that I can upload onto my GitLab and actually then deploy via NuNet.

So it’s super cool being able to do this, especially, you know, a lot of people ask or say, you know, the use cases are so vast, how can an ordinary person use it? If I can use it and I can deploy these things, I’m pretty sure that anyone can. And it’s really just being creative about thinking, okay, you know, one system I’ve done is like an email triage system, which is basically a little Python script that I have uploaded onto the appliance. And it just runs every day, or I have a daily report that runs every day that’s ran through the appliance.

So it’s kind of being creative and thinking, okay, if I’m using OpenClaw for this, you know, I can actually deploy it onto NuNet. I can even deploy with a private AI. So, you know, things are a little bit more secure and running it that way. So it’s pretty vast in terms of what you can do.

Yeah, it’s interesting people setting up their own OpenClaw environments on their computer that a lot of people don’t realise that you’re essentially opening up OpenClaw to everything on your computer. Yes. Having a VPS or tapping into a NuNet here gives you that level of isolation between your AI agent and your own personal data. I think that’s really important.

Something that people don’t think about. Yeah, yeah, exactly. Especially from the, you know, the security perspective, that’s why everyone’s running out now buying Mac minis to run these things on, which is actually one of the reasons why I switched over to Claude to look at that, because I can do it in a more secure way where, you know, my API keys aren’t accessed or pushed anywhere. Everything’s kind of local on my desktop.

But yeah, it’s definitely something that people should be aware about when you are deploying OpenClaw that it does have access into your computer. Therefore, you have to be super careful and secure about where you actually deploy the OpenClaw instance. Okay. So since we’re talking about security, can you give me an overview of how secure it is using NuNet?

Because we’re essentially using other people’s compute power. How do we ensure that those computers that are connected to the network don’t have access to our data, aren’t seeing stuff on there? How does that part of it work? Sure.

So NuNet is built on this zero trust model, security model. So that means that every machine has to be cryptographically identified and verified at every interaction. Right now, obviously we’ve just launched into mainnet or to live network. So we’re still like testing little things there.

But the premise is that this zero trust security layer means that we can securely access into the stuff that we need without accessing into the stuff that you may not want people to access into essentially. Yeah. I’ll put links in regards to that. And I think Kabir spoke about that in one of our previous podcasts.

So if you want to go back, I’ll put links in the top right hand corner for anyone that wants to get a better understanding of that to read up on. And it’s quite interesting stuff on how it’s all done. Now, what about costings? Because like Hostinger and VPS servers, they’re all fairly cheap.

How does NuNet solution here compare? Sure, sure. So I think that’s a very, very good question. And I think one of the things to note is that, you know, we’re not going to compare directly in comparison to AWS and Hostinger and everything like this, right?

We’ll be, you know, obviously looking at, and that’s kind of what we’re doing now is the experimental phase of like looking at, okay, how much actually does it cost to deploy these jobs? How much is it actually, you know, is the person going to get from their compute? So in terms of pricing, we are still like ironing out the tokenomics side of things. But one of the things that we actually have implemented is that the compute that you are renting or that you’re buying is actually based on stables.

So it’s based on USD. Therefore, you’re not basing it on NTX, which could go up and down 20%, 10%, whatever it may be over a couple of days, right? So it’s very hard to price compute in a token that fluctuates. So what we’ve actually done here is we’ve priced the compute in USD and that then translates, there’s an automatic conversion into NTX.

So therefore you’re not paying for, let’s just say one day it’s $10 and the next day it’s like 20, 30, or the next day it goes down, the person’s only getting 250 or something like that. So when it comes to that side of things, we are trying to be very fair with the pricing. Obviously as well, it depends on the specs, let’s just say from the compute provider side. If you’re somebody that has a older GPU rather than a newer GPU, how do we fairly price that?

So we’re still ironing out all of those kind of little bits that will, and this is kind of where we’re getting to in terms of when we launched Network Live, we mentioned that we were doing a seeding phase and this is part of that. So over time, we will have a kind of better idea in terms of how to price compute and we’ll have kind of algorithms in the background that will help to match, let’s just say your device to what somebody needs it for, as well as your reputation on the network, right? Because that’s obviously another big thing, is that if I onboard my compute or if I deploy a job on somebody’s compute and they take their machine off the network or they’re unreliable as a compute provider, the network will find another compute provider but it will also be kind of like maybe a different pricing section or it’d have to be, this is kind of where the orchestration layer comes in, is that they’ll have to find the compute to match that job.

So right now, like as I mentioned, we’re doing the seeding phase, we’re testing all of this out, we’re testing out the pricing dynamics, the tokenomics dynamics and everything like that, but I think the most important thing to mention is that all of the compute will be based on a stable currency. So something that we’re all used to rather than the the token side where it can fluctuate up and down quite easily. Okay Jen, talk me through how to actually use this platform. I’m pretty sure everyone that’s listening in already is pretty excited to utilize some of this tech.

Show me the platform, how do we get up and running and what’s the whole process look like for a user? Yeah, of course. So for those that may have been around from the start of NuNet when we first launched out the DMS, it was very, very tech heavy. It was very hard to get on board, you had to have a certain level of technical expertise to tap into the network.

So over the last year, we’ve developed out what we call the NuNet Appliance. And the NuNet Appliance is essentially the software and the gateway that lets anyone connect their computer or their compute into NuNet. So onboarding your CPU, GPU, RAM and start participating in the network. And then from the other side, it’s also the gateway where people can access into this network and to actually tap into this bare compute.

So to get started, you can head to docs.nunet.io and you will see here that there is a NuNet Appliance tab. Then you will just head down to installing the NuNet Appliance. Depending on your machine type, it will vary. So we have VMs, virtual machines, we have cloud InDescripts, we have ISO images and Raspberry Pi images.

So for me, I have a MacBook. Because it’s a Linux native piece of software, what I have to do is download a virtual box. And basically this is just our virtual machine and we’ve chosen VirtualBox for this. And this allows you to, in a simplistic way, it’s like having Linux on your PC, even though your PC isn’t a Linux PC.

So then from here, you will be able to download the VirtualBox, which we have the links to as well. Once it’s downloaded, then you will have two different images. And it depends on your MacBook type or your computer type. So I have a MacBook M1 chip, so I will choose the ARM64.

Then all I have to do is download this package, let it download, and then it will appear in my desktop or in my downloads folder as a little kind of square box, like a kind of rusty coloured square box. Then from there, when I double click it, it already automatically opens up into the VirtualBox manager. And then from here, there’s this one important step. We have to come into settings and come down to network and just make sure that it’s on bridge.

Usually it’s set to NAT, but we need to have it on the bridge adapter. And from here as well, you can just set it to NAT. And then we’ll be able to You can also, also mine’s already loaded, but you can basically change how much compute you want to contribute into the network, so from there, once you have that switched over, you will just click on start and then it will open up into a window like this, so there’s a QR code, there’s kind of a URL set up and a lot of different terminal looking side of things, but don’t worry, this is the only time you will look at a terminal within the NuNet appliance, so from here, what you can do is you can actually scan this QR code with your mobile phone and from there, you’ll be prompted to create a password for your account, now, when you open up and you create your password, you get a screen like this, so this is the NuNet appliance screen and as we can see from this screen, we’re just on the main dashboard here, we have our peer ID, we have our DID, the version of the DMS and we’ve actually recently in the last month or two, added this functionality into the appliance that you can automatically update the DMS, which is the device management service, which is basically the piece of binary code that NuNet is and also update the appliance version directly on the appliance, so beforehand, if we had a new version, we had to download, we had to, you know, go through the whole virtual box thing again, from this side, we don’t have to do it, we can just do it from one click, over here as well, you’ll see the ability to off-board, so I can off-board this device here or on-board your device, so this is really, really nice in terms of, I don’t have to, I can keep virtual box running in the background, if I want my compute resources, I can just off-board them really easy or if I want to on-board them, I can also do it from there, from here as well, we also have the wallet integration, so we have MetaMask and we have Eternal, of course, on Cardano, we also have a logout feature, we have a dark light feature as well and then down here in the status section, you will be able to see the status of your, so everything’s green here, we’re on-boarded, we’re not relayed, we’re running, we’re installed, some system information and then we have these three boxes here and we’ve just added this in, I think last week or the week before, where it’s an expanded details of your resources, so the free resources refers to how much resources you have free on the network right now, allocated resources are those resources that are being used currently for jobs or for workloads and then on-boarded resources are the total resources that you actually on-boarded from the network, so free resources plus allocated resources would equal your on-boarded resources, which is your full resources.

Then down here, we have like a Docker and basically what this shows is what’s running on your compute right now, so right now we have the Unicaddy proxy that’s running, but that could be anything, maybe somebody else has deployed on your network, maybe you’ve deployed on your own compute, whatever it may be, then we have connected peers, so these are the other peers that are on the network, so we have about 220 here now and that’s kind of the main dashboard side. Now to do a deployment, so this is like, as I mentioned, this is the nice and easy way to on-board and off-board your device, but the next thing is deployments and how we manage them essentially.

So the first thing as of yesterday, we’ve made joining an organization automatic and basically what this means is if I jump into organizations here, sorry, now my internet’s a bit slow, you can see that there’s two here, we have NuNet Network Live and we’ve NuTest Net, okay, and an organization is basically a governance entity with its own data and device management service context, right? So what this means is that if you want to deploy workloads and you want to create contracts within NuNet, you will have to join a certain organization. Now in this one we have NuNet, but the cool thing about this is that we can actually create organizations with other tokens, so Auki for an example is one of our use case partners and we’re working with them to help their community deploy nodes to run, basically they run robotics and they have a problem with latency.

Now they have a community that onboards their nodes onto the network, but it’s quite a long and lengthy process. What NuNet has done, we’ve come in with a POC and we’re continuing, sorry, to develop it out, but in that instance, they will create an organization within this network, okay, and onboarding their relay node will be as easy as just what we saw there in terms of onboard, offboard. And from this side of things as well, they will actually be able to have payment in their own token. So from the NuNet side, we’re actually not paying for that compute because that’s for the Auki community to do their own compute, you know, compute things essentially, but we actually will have a small charge for the orchestration.

So that’s where the NuNet fee would come in there, the role of NTX there. And that’s also another important thing to note that NTX isn’t just to pay for the compute, it’s also to pay for the orchestration fees. So by looking at these organization structures, imagine we have, you know, 15 DePIN projects that want to utilize NuNet’s orchestration layer, but obviously have their own token, have their own community, so on and so forth. What we can do is actually create up these organizations, allow their communities to join in, to deploy, to run whatever they need to run.

And it’s all handled in that project’s native token. And then the settlement layer in terms of NTX will be done on the orchestration layer. So this is kind of one of the things that we had envisioned when we were in the white paper is that we would kind of become a network of networks. So we’re not limited to like looking at other DePIN projects as our competitor.

We would actually look at them being people that we would like to work with because they can join into the network and it’s easy for their community to join in. We also have the orchestration layer, which is very lightweight in comparison to Kubernetes or anything like that. It’s a lot easier to use. That’s super exciting.

I wanted to understand organizations and that explanation was fantastic. And I could see that opening up so many opportunities for new DePIN projects out there. For sure. For sure.

And this is part of our plan to work with other DePIN projects is to make compute accessible for everyone. Not just those that have, you know, the technical capabilities or expertise, but actually make it easy for anyone to join onto the network, to join onto another DePIN network without it having to be a massive headache essentially from that side of things. So yeah, once you’re joined into an organization, then you will be given essentially rights to deploy jobs. So as you can see here, I’m in the NuNet network live.

My wallet is required of course, because we will be running jobs. And then from this side, it makes deploying really easy. So if I want to come to a deployments, well, let’s go one step back into ensembles. If I want to deploy anything on NuNet, I have to basically upload an ensemble.

And an ensemble is basically a template for a deployment in NuNet, right? So in NuNet, compute workloads are organised as compute ensembles. And a compute ensemble is basically a collection of nodes and allocations that work together to perform a distributed workload. Nodes in this case, obviously represent the physical or virtual hardware, and then the allocations represent the individual compute jobs.

So think as an ensemble of, I need this node, here’s the compute job, and put it into an ensemble file, which is a YAML file, and then find something on the network that essentially can do this. So that’s what we use ensembles for. Now, before I kind of actually got in to look at ensembles, I was petrified essentially of how the hell this could work, but when we actually break it down and we actually have it in our documentation here, an ensemble, which is all the information is here, we have a look here at the allocations. So the executor is going to be Docker.

The type is a task, and a task essentially just means it’s a once-off job. And the resources that are needed, so CPU, how many cores, GPUs, RAM, disk. Again, execution, and then the type is Docker, and then the image would be hello world, you guys can read more into this about YAML files and the structure of that, but that’s essentially what a YAML file does, right? So on the network, it’s just saying, okay, here’s the job I want to run.

Here’s the compute I need for it. Here’s a file that pushes them together and you can upload it onto the network. So an example of this would be something that I’m sure some of us are familiar with and which hopefully they do, this one here, so I can actually jump in and show you this here. So this is the YAML file for N8n, which allows you to self-host on N8n, right?

Which is like an automation platform. So here we can see it’s version one, the contracts, this is done automatic. So previously what we had to do was we had to get the, if I was deploying it, let’s say on your computer, I would have to get your contract did and your host did, right? Right now it’s automatic, so it scans the network, it finds it for you, finds the best compute for the job.

Then from here, we can see that the executor is Docker. This is a little bit different because it’s a service, it’s not a task. So a service is something that is ongoing. So if I’m running, you know, NuNet support on N8n, it’s not a once-off task.

It’s something that has to run all the time. So we have service in there. Then we have GPU cores, RAM size, disk size. Then we have the execution, which is Docker.

And then the image here is N8n. And then we have the environment variables, right? So anything that’s run in Docker or has a Docker image, you’ll be able to get the environment variables, which are here. Then we look at the type or the volume, which is local, the nodes, and then subnet join true.

Then we have a JSON file, which is basically just like here, like the name of the workflow, the description of what it actually does, this refers really into the automatic, like assigning of compute within NuNet, then we have the allocation and the default min, and we have some other things here, allocation resources, RAM size, disk size, peer ID, contract ID, contract host. Again, all of this is on, if people want to get deeper into it, it is on our docs site, so you can understand it a little bit more. And one of the things that I actually done to understand it is I just put it into Claude and I said, Hey Claude, I have no idea what a YAML file is in the context of NuNet, and I just shared the link to the docs.nunet.io site.

So it’s something like this, and it helped me to learn about how it’s set up. The cool thing about this as well is that, let me just cancel that. Cause we, we don’t need to change, change it. But the cool thing about this as well is that we’re working on building out templates so that you don’t have to have these YAML files, you don’t have to know these YAML files, right?

So n8n is a great example. So we can essentially deploy n8n on the NuNet organization. So anyone that joins into that organization will have that template there. And all you have to do then to deploy it is just click on deploy.

So you can deploy locally, which will be on my device right here. We can do a targeted deployment, which would be looking, let’s just say, Pete, I know your peer ID, I can type it in here, or a non-targeted deployment. And this is where the orchestration essentially would come in. This finds essentially, yeah, what it says on the tin here, the network decides where it’s posted.

For now, I can do a targeted one because I know I have this machine online. Hopefully that is the right date for that. Of course, sorry. I’ve tried a few n8n ones, so let me just have a look here.

Actually, I have one running. So this one here, basically I would have just deployed what I’ve done there. And then what I can do is just view the details and it’ll take a couple of seconds to load as my internet is quite slow, but you can see here that the status is running. Again, it’s a service, so it’s not just a once-off task.

From here, I can also shut down the deployment. I can refresh the information. We have the deployment progress, which here is running the manifest info. Then we have the allocations, the nodes, where it’s actually running.

We can see here that it’s actually running in Brussels in the EU. And then some new things that we have, which of course my internet’s not loading right at the moment, but we have added in DMS logs, right? And the DMS logs is a really nice way for you to get an overview as to what’s actually happening once you deploy a job. So once you deploy the job, you’re able to see in the logs of, okay, job submitted, found a provider, deploying job, basically all that kind of stuff.

And then if I want to access into n8n, I basically click the URL that’s here and it will open up into the n8n page and I can essentially sign in from that site and I can push in here, which I don’t think I remember the password for this appliance one, I do. So you can see here that we have some workflows. So this is the workflow that we created for the Discord to Slack integration, essentially. So this is running, taking a webhook from Discord and pushing the messages into Slack.

And this is running basically our support system so that we can make sure that anyone that has any issues on Discord, that they push straight into our Slack channels and that our team is alerted. And yeah, that’s kind of a very quick run through. Payment side, we have just here. So you can see here that we have some payments outstanding.

This is. As we’ve been testing, but it’s not been with a community machine, so they’re okay. But essentially what you would do is just click pay now, we can have a look at the ones that are paid here. As you can see, it’s based on USDT and then it’s converted.

I know my, my screen here, let me just make that a bit bigger. Converted to NTX at pay time, that’s how much USDT it would be worth. Payment details, the address, the blockchain, everything essentially you need from that. And then you can always view it here, which brings you into EtherScan or if on the Cardano network, it’ll bring you into which one it is on the Cardano network, but that’s essentially how, how things are kind of brand on the network.

There is more technical stuff that I haven’t got into yet, which is the file system, which I know the tech guys are working on that and have more understanding than me right now. I’m just on the, the basic, the basic layer of actually just being able to deploy jobs in an easy manner. I think that was a brilliant overview for anyone that is wanting to start using it and whether it’s an AI agent, SPO or even a web build, I could see, because we’ve got a web agency and we do a lot of automated builds that fire up servers in AWS or uses auto deployment servers and that costs us money. We have to have subscriptions for it.

So using something like this could be another alternative for that. So we’re not paying for all these different subscriptions for different third-party services to simply deploy, build and deploy a website. So I, I think this is great. This is brilliant.

Yeah, yeah, for sure. And the other thing is that I haven’t mentioned in this call yet, that NuNet is a fully open source project and we welcome anyone to, to come on and contribute to the code or to build on top of what we’ve already built. Like that’s the, that’s the goal essentially is it for it to be a self evolving network. So I’d invite anyone to, to come into GitLab, have a look at our code, contribute to the code, maybe even build your own ensembles and deployments.

So we can put them on the network for people to use. Like I was mentioning, it’s quite a vast, in terms of use cases, it’s quite vast. So we just saw n8n there. I’m now working on OpenClaw, OpenClaw private AI, LLM, which is Qwen, and actually deploying that on, on the network as well, so it’s quite vast.

We would ask anyone to, and especially with agentic AI, it’s popping up everywhere. So why not try it out on, on the NuNet network? Awesome. Okay.

Where can we find out more information? How can we contribute to the open source project? I’m assuming we can start at the docs, but how can we find out more and, and start using it? Yeah, for sure.

So everything you need to, to get started, of course, is on the docs.nunet.io. We have all the links and everything there. We have a link here to our GitLab. If you would like to contribute to the code, obviously this is us opening into the docs, but you can contribute to the code here.

If you have any trouble or issues, as mentioned, we do have our support, internal support system running on the network. So just jump into Discord. It’s discord.gg forward slash NuNet and we’ll help you out there. And for all latest news, what’s happening in NuNet, you can check us out on Medium and on X as well.

Brilliant. Okay. As always, links down below in the details description there, and also in the pinned comment, but Jennifer, thank you so much for joining me on the podcast, giving us an overview of where you guys are at and this demo as well on how to get started. It’s been brilliant.

No worries. Thanks a lot for having me on. Awesome. Thanks.