Swarm Community Call — October 2025

So hello hello everybody and welcome to Swarm community call October. For those of you tuning in on X, please note that uh this call is being broadcast from Huddle and this means that if you'd like to post a question or join the conversation later on, you're kindly invited to join Swarm's Discord server later on and uh join the conversation in stage chat or just drop us a message in stage chat while the call is ongoing and we'll pick it up from there and bring it here. uh as you know X is audio only for you but you can also I think post comments uh in X and we'll try to pick them up and answer live as the call goes. Um in today's October's call the agenda is as follows. First, as per usual, we'll start with core development updates and next we'll move to announcements dedicated entirely to the upcoming dev connect in Buenosiris uh where the team is heading to and after that we'll move to the infocus section where you'll witness the live launch of Bport an on-ramp app to swarm and in the end chart will explain how AI can help devs work with swarm.

All right, let's kick things off. Uh, I'm welcoming first Nikki and Callum to do to do the core dev updates. Hello Nikki and hello Callum.

Hello.

Hi. We can we can hear you both. Excellent. The mic is yours.

Amazing. Amazing. So, I'm going to kick start. So, hello Mina. Hello everyone.

October uh Storm Community Call. Great to see you all again. This month we've been fully focused on finalizing the B2.7 release and I'm happy to say things start coming really well together. Uh let's begin with engineering team that has been fully focused on getting 2.

7 ready. Most of the plan work is complete and we are now on the final fixes review and testing phase. One of the biggest pieces this month was multiple underlay address support. This allows nodes to advertise more than one reachable address important for complex networking setups and big in the browser. Implementing this um led us to clean up related parts of the code base which improved the overall quality and clarity of B always keeping the scope of our objectives protected.

Um at the same time both autotls and shock replica have been completed and are now in the final testing phase. Uh autotls enables node specifically uh browser based on uh to establish secure connections automatically and equally soft replica improves uh data persistence by introducing uh replication for single owner chunks. Uh we've also improved our CI/CD pipeline uh adding more test cases and strengthening our current uh automated quality checks, meaning future releases will be more reliable uh out of the gate. And with all other 2.7 issues resolved, we are now down to just two last uh items uh autotls and under layer address.

Uh and we are close. Um regarding uh the research team uh two key areas in focus staking and incentives model uh are being uh reimagined with updates planned around changes to withdrawal stake and upgradability ensuring the system can invol can evolve more uh easily through releases. Um and also the long-term plan for neighborhood splitting and rebalancing is close and this is an important step uh forward improving network uh robustness scalability as the number of uh nodes uh grows um and kum will elaborate uh more on uh these items. So wrapping up um 2.7 is nearly there with shipping new features and uh raising the quality bar across the project.

Thanks again everyone for testing running nodes or just staying engaged with us. Your support keeps moving forward and see you next month with more updates. Thanks a lot and stay tuned. See you next time. Back to you Mina.

Thanks Nikki. Thanks you also. Please stay tuned in case we had some questions and um you need to be here to answer them. Um okay. Yeah.

Callum, would you like to um add anything at this point? Just to

Yeah, just basically to sum that up of what we're up to at the minute. So with a 2.7 coming up that's kind of going to pave the way for our in browser client to get to, you know, this proof of concept that we can get out and get working with everyone. And to sum that up with what's coming next on this call, which is, you know, an introduction of where is at. So really what's happening with these releases and where org is moving in a in a pro is to this completely usable product layer on top of things where we're able to upload, retrieve, view, download anything without actually having to run a node yourself as well.

So I'm very excited for what we've done so far, this release upcoming, and the products that are going to come next. me as well I have to say and def connect is approaching and with that said um in the next in focus section you'll hear or like Aaron will walk us through the um Bport app but before let's uh just uh say a few words on this dev connect because Swarm team members will be there so reach out to us on one of our socials to connect or catch us on the spot we'll be uh kicking off by attending a pre-event side event by web free privacy now. It's the second cipher punk congress which is on the 16th of November. So catch us there or later during dev connect to either talk about anything that you've hear or you will see in the later on demo today or anything else really. Um solar punk team members will be hosting an amazingly interesting side event themselves on the 21st and I would like to kindly invite anybody who's going to dev connect to also attend that.

is going to be in Club Lucero in 10 minutes walk from main venue and it's starting at 4. Um you will be able to witness a round table on decentralized AI moderated by Luke Kernner from Crypto Mondays with Shandur Swarm COO Chakin speaking for data fund and D AAA members uh and Michael Canour from Hashgraph. Uh you'll also be able to uh attend a workshop uh titled launch your own AI agent on swarm and uh hear a presentation and a demo on how to decentralized any front end and we'll be wrapping up or they'll be wrapping up with a deep dive workshop on multimedia streaming on swarm. So, it's going to be a really exciting uh evening with Solar Punk team in Argentina on the 21st of November. All right, that said, uh I would like to welcome Aaron on stage now to officially launch Bport.

Hi. Yes, thank you. Uh sound check, please.

Oh, yeah. Yeah, you're everything's nice. I can.

So, let me share my screen. And can you please also confirm that now you are seeing a Bport text in

and now you are seeing a Bport text. Yes,

that sounds right. Thank you.

All right, let's go. This is super exciting to to finally see this app in in action.

Yeah. So I'm going to talk about Bport. But uh before we do that, I would like to travel back in time a bit to the beginning of this year and talk a bit about what options there were to upload something to Swarm. So on the left you see running your own B node and I'm going to talk about issues here. So in the beginning of this year we had some issues with running your own B node.

One of them was the long syncing times. That meant when you started your B node for the first time, it uh took about 40 minutes to synchronize from the blockchain and that happened even after you managed to fight your way through the documentation. And the other issue with that was theis funding. So you not only had to have BZZ, you had to have XBZ on the chain. But undeniably running your own B node is the decentralized approach.

It is always the primary recommended approach. It's also healthy for the network to run your own B node. Now the other option was and it wasn't really a good option was using this warm gateway but uh it had very obvious issues like only small file sizes were supported. uh the duration of uploads were unpredictable and it was guaranteed to be a short time span but still we could see that this was a very convenient way to upload content on swarm just go to a page select your files upload and without running uh your bnode you had them on the network after seeing the issues that I mentioned before we had a plan for this year to improve the UX in these areas as so on the left as part of the node spin up UX uh improvements uh we embedded the postage snapshot to the uh B binary which got rid of that 40inut uh synchronization time uh upon the first launch now you can run your own B node in like under 5 minutes we also got rid of some other artificial waiting time that was for stabilization it made it even faster to be able to start and use your B node. So these items are delivered.

Now to address the second problem with uh running B nodes, this nosis lock in or the inconvenience of getting XZX die bridging them somehow over to Nosis. We were working on projects related to multi-chain funding or multi-chain payments. This is uh currently in progress. Two projects under this umbrella are a so-called multi-chain widget that can be used anywhere. It can be either statically hosted or embedded in applications.

And this purpose is to make it easier to fund your own B node and get get both XTI and XBZ. This is again to promote the decentralization of this warm network to make it not only fast to run your B node but also make it have good UX uh by eliminating the funding issues. Now on the other side uh and we are here for this today is Bport. So Bport is a standalone application. You can think of it as Swarm Gateway but on steroids.

So with B for you don't need to run a node uh you only need a wallet uh some funds from many uh Ethereum compatible chains and it requires the least effort and it's the fastest way to upload to the network. So again you can see that uh on the left running your own B node the decentralized option and on the right the convenient option and this brings us to Bport starting with a uh screenshot on how the main screen looks like and we are going to explore the use cases uh through several small clips. So the first one what which we are going to do is purchasing storage uh through bport and in this main form you can select the chain that you want to use the token that you want to use the storage capacity and the storage duration. You get a quote for that that you can execute. Uh MetaMask pops up or your wallet of choice.

Uh you confirm the transaction and then a spinner starts. uh that deposits your funds waits for transaction confirmation and starts creating the the postage batch. Once this spinner finishes, you get to this popup where you can already upload to the swarm network. So in this first use case, we are going to start simple and then and just upload a single file. So I'm selecting a PDF.

Clicking upload get a signature request so we can authorize ourselves for the proxy uh which is between our browser and bport. And just like that it was already uploaded. I got a link to a gateway and you can see that uh the file loads and is visible on the network. So let's get a bit more complex than that. Uh this time I'm navigating to the upload tab where my previously purchased batches are displayed.

So I'm selecting the one that I just bought. And this time I'm uploading a folder. So I'm here uploading a folder uh with PNG images flags of the word confirming um the signature request. Upload is already done. And when you upload a folder a folder index is created.

So this lists all the files that have been uploaded in that folder. Uh you can scroll through them, click on them and they load on demand. Of course for the third use case, let's do something even more advanced and that is uploading a application or a website. So I'm doing the same flow as before. I'm doing a folder upload but this time the folder that I'm uploading is the source of a website.

So in its structure there is an index html uh that that makes it very obvious for bport that it's a website and it will upload it like that. Now having done that I'm going to navigate to my history where I see my latest upload and there I can click the ENS button uh for further features uh related to ENS integration. So what I'm going to do now without leaving Bport is registering a new uh ENS domain. I'm going with flagg guesser and starting the registration process. Uh so with the magic of video editing uh we are going to cut some waiting times uh shorter.

For example, the first step to ENS registration is always a one minute timer. That is in order to avoid the front running. Bport explains these steps in an easily understandable way and most of the time you just need to uh click the primary button to proceed to the next steps. This is the actual step where the ENS domain registration happens. Uh gas fee is a bit high on Ethereum at the moment, but you can expect that.

[snorts] And after some more waiting, just a minute or two uh within the application, you get the confirmation that your ENS name has been successfully registered. So, this is not the last step. As the last step, you still need to link uh your swarm reference, your upload to your ENS. So to do that, you go back to the send set content hash tab and there you select uh the ENS name that you want to use. Simply again just click the primary button.

You get the transaction request for setting content hash for your ENS. And as that transaction goes through, you receive multiple options to access your content. So I'm picking E limo here. And once I navigate to that tab, you see that this content is served uh from swarm and it was fully uploaded uh through Bport. So this is just a uh silly flagg guesser game that I made where I accidentally got three out of three.

So yeah, again we have seen that now uh there are multiple options to upload just a single file, a folder or a website and then even go a step further and then set it up uh with ENS uh giving you multiple options to access your content. And I want to emphasize that even though Bport is a centralized component to get your data on the Swarm network, once it is on Swarm, that content is still decentralized. Uh the URL is now uh displayed on the screen. I welcome everyone to start using Bport and please give us feedback either on Discord or using the feedback button uh within Bport for issues or ideas that you may have. And with that, I would like to thank uh you for your attention.

Amazing. This looks so so awesome. I mean obviously I have uh done several uploads already and I do also invite everybody uh listening to this or seeing the link later to yeah go explore it's super simple compared to how and what decentralized storageages usually mean. So no node running whichever token that's Ethereum or like EVM compatible works and uh yeah you get your content or full websites up in basically no time I would say. So yeah thanks Aaron for this super comprehensible demonstration.

Um I'll be checking if you have any questions in Discord. Uh I will also try to check in X. Um, but while I do that, I already invite CH to the stage for our community talk section. I hope CH we can hear you now.

Can you hear me, Mina?

Oh, yes. I was kind of worried because your your avatar looks panic like he's panicking, but

yeah. Yeah, he was panicking a bit beforehand. So, um yeah, thank you. Great that you hear me. A bit later I will check if you um can see the screen that I will share but um maybe I'll just start with a bit of an intro.

So um we'll have a short demo on how AI could be used to support development on swarm. The paradigm of software development is changing. So no longer do you look at the documentation of a product. You use AI to either look it up for you or even code it for you directly. So in that sense to onboard developers you not only have to make the docs available to the developers in human readable form but you have to make it available and known to the artificial intelligence.

Different models are already a aware of how swarm works to a certain degree but it depends on when they were built and how they included the info from swarm documentation that they might take the wrong or obsolete sources as the truth. So um also mostly they're able to browse web pages and could look for the latest version of the docs online. That's certainly possible. But perhaps a more elegant and currently free solution is to use an NCP server that indexes the docs and just returns the relevant parts when you query it. Um so saving tokens and time.

Such a tool is context 7 which you can I will share later in discord the links to um more information about this um but maybe at this point I will try to enable uh I mean I will screen share please uh can you confirm that you can see my screen uh Mina

yes I can

okay thank Thank you. Thank you. So um so a developer would probably use something like cloud code, Gemini CLI or Codex in their workflow. So let's look at cloud code in this case. Um obviously you need to have cloud code installed as a prerequisite.

So um and also add um this MCP plugin. I'll share more info in the discord chat. Um and um then you go to the to the folder with your project, right? And you run cloud. So cloud says hi.

And um then maybe first thing you would do is uh run the command. So /mcp to check if the server is indeed uh the MCP connection is indeed established. So you can save this to context 7 and um then uh you you run a query you you ask it a question and you specifically instruct it to use context 7 so that it doesn't use some old information uh somewhere in its model. Uh and it's also best if you specifically ask it to um to use a specific library index in there so that things are faster, right? So you know this would be the optimal way.

So I tell it to use context 7 and /hatterphere/box library and tell me how to run my b node in a couple of paragraphs. So uh now it's thinking and um for any tool use uh you have to approve the use or just approve uh uh let's just say always approve or something like that if you don't want to do that every time and um it's doing a bit uh of querying and it will come up uh hopefully very soon with a couple paragraph description. So um there it is. So um we're not going to read it but you can also you know ask for a more detailed description. So on this is just for um demo purposes.

So let's maybe do another example. So let's ask it to write some code right uh draft the typescript function that will assume a b node is running locally on the standard port and we'll connect it to it and upload a file a user selects in the browser use context 7 and the specific library again and it's zigzagging a bit. Um so why use context 7? because um it uh has sort of almost the latest docs. They were indexed a week ago.

Um so uh I don't think there are much changes since since then. So I think um and you can al also reindex them on demand, right? So that's uh when changes changes arrive, you can uh you can do that on the context 7 page. So um this would be the function um and some instructions with it. So um and um this is it for the short demo.

So your imagination is the limit here. Of course note that you will rarely get code that does exactly what you want on the first try and the iteration is the key. It might also be worthwhile to specify um some tests in advance and then have it iterate until the tests pass. But that is um um not in scope for today. So uh call to action I invite you to use some discord to share what kind of workflow we use as a developer and maybe these tools fit into it what else we would need and thank you and back to you.

Well, thank you ch for this. Absolutely. So, how just a question um what's your experience? So, is this is this a reliable thing if you ask him to you know code for you because we are hearing a lot from developers as well that yeah it's uh you know on on the actual usefulness of AI tools at this current stage where they are. So, are they u reliable or not?

Yeah, I mean you wouldn't just, you know, commit this to as a PR, right? You would check if it works and so on. And as I said, it would probably not work on the first try, but it's like you can then talk to it, right? You know, this this doesn't work. And if it apologizes, obviously, oh yeah, how could I miss that?

And then eventually, I think it depends on the task, but uh um eventually it will uh you know, converge to the to the solution. So if it's faster than you would do it uh I don't know I guess if uh I'm not a developer so it's definitely faster for me to use AI than to study everything. Uh but um it depends I guess right I definitely I would um yeah I would say it's very very good from uh my opinion how it works. Anybody else here in the call with experience? I don't know.

Aaron, did you ever try and use it or is it like because to to me it sounds like with all AI it's really it is super fast and useful but it's as good as you are you know guiding it or you know actually having some knowledge up front as to what you want to get like you need to be able to check it right.

Uh yeah I can give my thoughts. So I've been using tab completions for years now. I'm using copilot for that and that is a super useful feature. I do use code from time to time but it's mostly for tasks which are super common and don't have deep uh domain knowledge requirements like the deeper the domain is and the more um more it's custom to business logic the more mistakes the AI is going to make. But if I ask it to create a let's say custom react component which is very basic there are tons of examples in the uh training data that was used for these models then it pro it almost always oneshots the implementations and it's what I expect.

Thanks. Yeah, thanks for sharing. I I I guessed you would have something like that to share. So I'm checking the channels. Um I do see SA posting in uh Discord and I would like to invite you all to post your questions in Discord state chat please because I do hear that um in X you need to the to have a pro account to be able to comment from yeah from what I'm seeing in the message box right now.

Um and SA is also asking if yes we're if we're going to dev connect and if anybody here would be up for doing a stream to swarm podcasting setup with me. Absolutely. Um we probably will establish a def connect like a separate channel in discord but you can please just feel free to post more questions and reach out in state chat for the moment for as long as we don't have like a dedicated channel for defcon connect but the team will be there we can chat stream to swarm absolutely and um I bet that solar punk team would also be very willing and helpful with uh any questions when it comes to multimedia streaming to swarm. Um I don't see any other questions. So unless uh anybody would like to have some concluding thoughts uh I will just invite everybody to rejoin us for the next community call in November.

I think yeah the last Thursday of the month as always that falls on the 27th of November. will be back from Dev Connect by then and I guess the call will be very Dev Connect flavored. Um yeah, with that I thank you all for being with us and see you next month. which is um