AI Chatbots Revolutionizing Web3 Communication: Unveiling the Power of Relay!

This is the PoolTogether Community Podcast. PoolTogether is the world's number one no-loss prize savings account. You can visit pooltogether.com to deposit. You're listening to the PoolTogether Community Podcast. I'm your host, Millie Yearley, AKA Tim, AKA I ran out of gas for the second time in my life today, right before this podcast. But everything's okay. And my hands smell like gasoline. And I'm so happy to be here with Sean from relay.cc. Sean, thanks for being here.

Great. Glad to be here. Thanks, Millie.

Cool. Cool. Yeah. So my name is Millie Yearley because I'm manifesting that Pull Together will have million-dollar prizes, uh, at least once a year. So I changed my name to the thing I want most in, in, in life. Well, really for PoolTogether in my life. And so that's why I'm that way. But yes, you can call me Millie. Go for it.

Billy next year?

You know it, you know it. It was actually Millie Monthly is what I started. And people are like, Tim, you need to be a little bit more realistic. So I was like, okay, so I'll be Millie Yearly, but Billy Yearly will be next. Yes. And I mean, come on. I mean, that's, that's critical mass right there. If you have a billion dollar prize, you just gotta be deposited in PoolTogether. But we're not here to talk about Pull Together, Sean. We're here to talk about you, your life, and what's going on with Relay. So could you just give me your origin story? And then also, like, how did Relay start? And then just kind of explain Relay to us because it's really— you're at the hotspot of crypto and AI. So anyways, go for it.

Yeah, the story goes back a little ways. I first used Bitcoin. I was at Berkeley, the Bay Area, in 2012. Worked in the Bay Area for a couple years and then went to a meetup where it was this new cryptocurrency that was going to change the world. 10 people meeting up at a bar in Oakland talking about how XRP was the next big thing. And 2 months later, I went to an Ethereum meetup and I was like, oh, this is the one. So learned about Ethereum in 2015 when basically I was working in the Bay Area. I quit my job in 2017 to learn Web3 programming. And then went to the first ETH Global Hackathon in Waterloo. Actually, before it was ETH Global, it was just a hack at the first Ethereum hackathon, and I did a hack around MakerDAO and their new idea for a stablecoin and joined Maker after that. So that's how I first got started in—

That's awesome. I just went to the ETH Global Hackathon in Tokyo, and that was Super cush. It was very fun. And being a non-technical person like myself, it was cool that I got to actually contribute. I felt like I actually contributed even though I don't code, you know, but I'm trying to get dangerous now. Okay. So, so you, you joined Maker and then how, how did Relay come about?

So I was at Maker, then I was at Infura at ConsenSys, and then basically in This is now November, October, November 2021. For the past 4 years, I'd accrued a ton of governance tokens just from airdrops, the Uni airdrop, 1inch, ENS, Gitcoin, and I was trying to be in governance. And what I saw as a critical problem is that there's no way to communicate as your Web3 identity, your Web3 persona. Everyone is dropping to use Discord or Telegram for very DAO-related communications when Basically, your address can encrypt and sign a message. It can also— sorry, it can message. It can also sign a text message and encrypt it. And so, we started to work on this problem and I started Relay. It was actually called DAO Panel first. It was a control panel for your DAOs. And so, we started to work on that aspect of it, more DAO-focused and more messaging-focused. And then, this is now April of last year. XMTP came out with their beta and we're like, holy shit, this is kind of like what we're building towards. Let's use it right away and skip the next 5 months of development and go from there. So we were one of the first partners of XMTP. Basically last year we were building front ends on top of XMTP and that led to basically something like Intercom, but Web3, some Telegram, but Web3. And we started to build this group of companies and users that use their ENS name to communicate. And that was what related for last year. And the best thing about working in the chat space, November 30th, ChatGPT dropped and everyone was like, holy crap, this is pretty powerful. This is just a game changer. And in the past 6 months, we've seen exactly how much. That's changed the narrative on AI.

That's awesome. I just, Darby, I just sent you a message through relay.cc. So if you want to please get back to me, it's end-to-end encrypted, so don't worry, the feds can't listen.

Yeah, for anyone who hasn't heard of XMTP, it's encrypted, and then it's very much like Signal where you just have to create an identity once, and that kind of gets you on the network, and you can receive messages encrypted. So we found that great to use, but what we started to see is businesses wanting to create ENS names or even subdomains. So we've worked closely with like ENS DAO and this to onboard businesses where they can create, for example, chat.metamask.eth or chat.coinbase.eth. Like, doesn't exist yet, but you can imagine it could. And we were onboarding businesses and the problem at the time was businesses would have to have someone there on the other end to do customer support. You have to When Yelp introduced messages, now suddenly businesses had to have someone answering their Yelp messages. And this is at the same exact time that ChatGPT let people, let businesses create their own, at the time, maybe subpar customer support. But I think this year has shown, next year, that it's going to be above par, the average customer support that a startup can get. And it doesn't stop there. There's— I can go into the different Or I will go into the different use cases of why a Relay chatbot, which is Web3 enabled, is much more than just a— but that's basically how we kind of pivoted from being front ends for messaging to now we're creating bots that have ENS names and message through XMTP.

And there's no human, you know, support or addition to that, right? I mean, so, so basically you realize that you're talking to a bot that has the docs and the, you know, the dev docs and the white paper and everything fed into it. And then you're basically, you know, searching those as you're chatting with the agent, correct?

Correct. It's a process called embedding, where you basically take the base GPT-4, you embed new documents into it. It's a very interesting area of math called vector, like the vector space of text and comparing it to the documentation. And so that lets you basically add new knowledge to GPT-4. Very commonly is documentation, GitHub, basically the things you mentioned. So we add that to the bot and then that chatbot is able to answer questions about the business. There actually is some amount of human transparency where we bridge the XP Chat into Discord. And so there's human oversight over what the bot is saying and actually Human intervention if needed, so you can kind of take over when that has gone bad. A lot of the work we've done in the past three months has been you may have heard of like hallucinations where ChatGPT will very confidently lie to you and just tell you information. There's basically workarounds to get around that, and that's what we've been doing. The other area of kind of research has been how the bot can help with on-chain transactions. So. We basically had demos the past two months about a Uniswap trade and also an ENS registration where you can chat with the bot like, "I want to buy ten ApeCoin. I want to sell ten Uni." Whatever, and it'll just populate that for you and allow you to sign MetaMask. So that's the first step where the bot assists you with transactions. The next step beyond that would be giving the bot some allowance: one thousand DAI, ten thousand USDC, whatever. basically how much you're comfortable trusting this chatbot. And over time, we're going to be proving and building up that user trust in AI bots that can control crypto.

Okay. So that's 2 different audiences there. I assume like right now it would be more— so you asked, it's 3 different audiences. So you started with messaging on web, like XMTP. Am I saying it right? Yes. XTMP. X. Yes. Something like that, that messaging. And then you're, and then you're going into this new product, which is ChatGPT for Web3, but specifically like DevRel, you know, maybe investors could also dabble in it, but just kind of like looking at each project and asking questions of it and getting kind of that intel from the chatbot. But now you're saying like, oh no, no, we're, we're not going to stop there. We're going to, you know, empower a, basically a no-code Crypto bot, you know, that can do your investing for you, or it can maybe even do MEV for you if you want to do that, or could do, I mean, whatever, whatever you want to do on-chain, NFT collateralization. But so that's an interesting way to go about it. Why, why that path? Like, I understand, I understand the messaging and I understand like, okay, well, no one wants to answer messages, customer service messages. So let's put a bot on it and then yes, let's allow them to take over. So then in my mind, I'm like, okay, the intercom of Web3. And now you're saying, oh no, no, no, we want to actually go in this direction, which is more investor, you know, on-chain swapping, that sort of thing. How'd you get to that next frontier?

I personally don't have a background in the investor or the, I guess you call it the finance, the trading side of crypto. I've been, my trades have been a sell signal. So like, I wouldn't say our bots are going to help as a trading bot. It's more helping as For example, with this is actually not that far-fetched, but a pool together bot could help like educate you about what pool together is and tell you the different options you have with your stablecoins or basically what the pools are and how to actually do a certain basically advise you what they are and then could help you do a transaction. So we're thinking of it as an alternate front end to many of the DeFi crypto. websites out there, all of a sudden, instead of doing a, creating a custom front end, this is becoming a more, it can be a more user-focused way where like in the middle of a transaction, the user can say like, oh, okay, what, what is the variable again? What am I doing here? And the bot can take a second to explain and basically give her, give the user peace of mind on what they're doing.

Oh, okay. So it's like you're like on-chain chores. It's actually a different approach to abstraction. Where you could just be like, listen, I want this NFT. How do I get it? What are my options? And the OpenSea bot would be like, okay, well, this is what you have on secondary. You know, they might be doing a mint on something. But yeah, for PoolTogether, it's like, hey, I have this here. Here's my wallets. Here are my wallets. And like, I want to deposit into PoolTogether. What's the best way to do that? Okay. Well, it looks like you have X on Optimism. You can deposit that and you'll, you'll probably have the best odds. Like, that's the dream. I love that. And then I could just use my voice and be like, computer, or I love Alexa, but Alexa is not— the only thing I do with Alexa is, you know, ask like sports questions or turn on my lights. And so—

Yeah, like a Whisper protocol that I think is better than Alexa at voice recognition. And so anything you can do through a text conversation, you'll be able to do through voice. On that notion of OpenSea, there's Seaport contracts, there's ways the bot could assist with either collectors or obviously either sellers of NFTs or collectors trying to buy NFTs. So a question I want to dive into is how does a bot handle money? How does ChatGPT handle money? And there's some banking APIs out there. I've seen one startup trying to do one-time credit card payments to pay for AI services. But there's a— on the Bankless podcast a few weeks ago, they had on the Alliance DAO, who wrote a really excellent blog post on the intersection of AI and Web3. And there's a— half an hour into the podcast, there's a sort of an epiphany from Ryan where he's like, holy shit, this whole time we've just been baking— we've been making a financial system and banking rails for the robots.

There you go. That's what it is. And then you just, you just talk to them. You say please and thank you. You be polite with the robots, you know, don't just boss them around. And then, and then they can, they can do so. I mean, I mean, task number one is like, hey, can you do this transaction when gas is low, please? Like, that's one. I would love that.

Transaction when gas is low.

Yeah. What are some other, what are some other go-to, like, killer apps? Say that again.

Like, just morning question. It's like, all right, what are the kind of like urine, but like, what are the top 3 stablecoin yields today? It's like, all right, put my stables into a mix of those 3. Yeah. Then the idea of a chatbot, like one story, like chatbots can do a lot more than what we currently use them for, which is like basically withdrawing information from the chatbot. It's like a one-way flow of information from the chat to you. So when— why we're using XRP is because by nature you have to have your wallet connected. Like, your wallet is verified if you're chatting to the bot, and so the bot knows your token holdings, your NFT holdings, your transaction history, and the bot can customize the conversation to your wallet. So that's like a bidirectional flow of information. And we sort of talked about how there's actually a bidirectional flow of tokens that's possible. You can either send tokens to the bot for basically buy something from the bot in some cases, or the bot could use your conversation as a— there'd be reasons why the bot would distribute, say, a POAP or NFTs or tokens to you based on the conversation. And so this bidirectional flow of tokens is something that I really haven't seen anyone else doing, where there's a ton of features that flow out from that.

Oh, that's really cool. So like, I mean, this, this could also be like referral-based. Yeah, a little bit though. There's like a privacy concern, I guess, is like if you have the bot listening, I don't know. I mean, you can address that in a myriad of ways, but, but yeah, I mean, like for example, with PoolTogether, one of the things is like onboarding people to crypto is like such a necessary thing, but the most impossible thing to do, especially in this economy when the market's down and, and, you know, there's not the financial get-rich-quick You know, motivation for people to figure out crypto and just like make it work and figure out the on-ramps and stuff. But I mean, I could see that. I mean, this could definitely be just like you said, if there's rails or just like Ryan or David said on Bankless, like just, just like there are rails that we're giving AI for banking, like there's also rails into crypto, you know, depending on regulation and how, how that can work. But it's, you know, I mean, I already have it in my Venmo. I got my Venmo rewards turned into ETH. I'm just like a hop, skip, and a jump away from NFTs and, and, you know, layer 2s right there. I just need, you know, Elizabeth Warren to tighten up or something.

Well, it's—

well, we'll have to edit that out.

There's a—

I don't want to attack Elizabeth Warren on the podcast. Sorry. Go, go for it, Sean.

I'm in the US and right now I'm in, I'm in the EU, but The EU, the US, they're definitely making laws around both crypto and AI that may make AI transfer or like AI assistance of DeFi. Anyway, it's not going to delve into that too much, but what you're saying about privacy, I think it's like when you connect to a site, anytime you sign into MetaMask on a site, like they know your wallet address and so they know anything the bot would know about your token holdings and transaction history. The bot, the websites just currently don't really. Act on that. Yeah.

I think it's just more for the conversation, like, like, you know, the content of me messaging Darby through XMTP, you know, if I say some buzzwords or like, I think that's what you were talking about, right, Sean? You were, you were saying like, well, you could like have some sort of achievement and then get tokens based on your interactions with people. Like the, I mean, and then my mind went to like shilling PoolTogether. And so, you know, if I shilled PoolTogether enough in my messages, then I would. get pool tokens or, you know, you could set up something like that.

Well, XMPP is end-to-end encrypted. And that's the case, like when you're talking to a company, there's obviously some like terms where the company has access to your chat with the bot, but otherwise it's end-to-end encrypted and there's not, there's not currently group chat in XMPP, but they're adding that. So yeah, so I think we went over some of the use cases of why either basically a person— so, all right, sorry, I got detracked, but the example I want to share was something like, I think it was either 0xDiner or someone on Twitter where they shared the idea of a calendar or sorry, a Calendly where in order to book time with me, you'd have to convince my bot that it was worthwhile.

Yes. And I told them to delete that. I was like, please delete this.

Yeah. Like, well, like that concept of a bot that could be gating either free mints or user access to certain parts of the site based on completing knowledge tasks. Like what I was referring to is more like quests where the bot could ask for and prove certain knowledge from a new user and be like, all right, you've got this NFT now, you're part of the club.

Yeah, no, sorry. The one I asked to delete was the one where it said like how I think it was like, uh, how, how late you are to meetings. That was the one with a calendar. But no, I also saw the other one, like, just like how much of your time, like based on your salary, is it costing you? Which I don't like because we're human beings. We're not like numbers, but I understand business case or whatever. But no, I mean, I love this stuff. I love AI because to me it's like, um, little bits. They're like these little circuits that you can put together and they're made for kids, but I love using them because I don't code and I don't make electronics or, you know, and so it kind of like, but what if I could and what could I do with it? And so there's this really cool thing of like being a normie in that world, but then kind of getting dangerous with it, getting the tools, you know, having a template to be able to like, what can I do with crypto plus AI? And so just as we have abstraction for users getting into crypto and getting into and using crypto, I think like this whole like crypto AI space, I just love thinking about like, what are the possible uses? Like, what are some— and I know you've given me some, Sean, but, but I mean, I would love to see like a hackathon just on AI crypto use cases, even with just bots. And like you said, we need to get rid— we need to mitigate the hallucinations because those things are real and they happen all the time and it discredits bots, for sure. My whole synthesizer community does not like my Eurorack synthesizer bot that I made because it, because it lies half the time. There's that, there's the finances, right? Like you have to, it, not only can it not lie to you, but it also needs to make wise decisions. And I think after you have a few, a couple of people getting rich off of this, then I think people will be like, it's fine. But I understand. And then what other blockers are there to adoption or, or do you see to what else needs to happen before this is like

I think there has to be more specialized independent AIs. So I was at Zuzalu this month for this AI x blockchain, like a 2-day summit where a lot of the talk was on actually around kind of on the 10, 20-year, in some cases some people it's like 2 years out, but like just how world-changing, like very intelligent AIs can be. But we focused on more the next year or 2 out where there are— say every company has an agent doing things beneficial to that company. Every person has an agent doing things beneficial to them. And so these agents are on 24/7. And so you'll see less human-AI interactions and more agent-to-agent interactions. But going back to your question, I think proving out the case for the human-AI connections and how AIs can really assist humans with onboarding to crypto and using crypto is the goal of Relay. And I think that's a very tangible goal. We've seen benefits. We're working on some pretty major Web3 companies having very sophisticated DeFi chatbots that are going to prove this concept. And then once companies realize that they're kind of leaving money on the table or leaving customers on the table, That's when we'll see more of the flywheel of adoption turn. And then there's already a number of companies working towards this idea of an agent economy where you have these 24/7 agents out there that can kind of have a loop inside their heads of planning, prioritization, strategy, reflection, and they're kind of just acting on behalf of you. All the time and interacting. And so my thesis is that it's going to happen on Ethereum as the financial rails. It's going to use XMTP and wallet messaging frameworks for the communication rails. And people bring their own data and they bring their own prompts. That could be private data from Ocean Protocol. It could be private company proprietary data. And then Relay, we've done some prompt engineering. There's some new papers One really cool one is called Tree of Thought. We don't think there's going to be thousands, if not tens of thousands, of the smartest people in the world trying to create better AIs. There's the data side of it, there's the prompt side of it, and then there's the communication side of it. We're working towards a future where people bring their own data, they bring their own prompts, and we wrap it up into what we call a robot and put it onto the Ethereum and XMTP networks. And so that's how we get to this fully functional agent economy. And the reasons, I think the benefits why we do that is a lot more, I guess, just financialization of certain products, more efficiency, and sort of happier people that have less, more time for creative tasks when, say, your business-focused tasks or your maintenance tasks, your day-to-day BS that you have to deal with is handled by your own personal AI. That's basically where we're headed.

I love that. And it's an honest take on AI because right now I'm seeing a lot of Twitter influencers are like, use AI to create the best resume or cover letter. And then on the other side of it, it's like, use AI to summarize everything for you. And so it's like people using AI to put in bullet points and get out like massive texts, massive fancy blocks of text. And then people taking those fancy blocks of text, not even looking at them, putting them into the AI and getting back out the bullet points. So it's like, let's just, let's just live with bullet points, BB. Like, why don't we just, just share your prompt, share your prompt. And then, you know, that, that's all we need. So the other thing though is this, and when you say this agent economy, I think of devs because I'm doing that hackathon and I did, my team had 2 devs on it and everybody else was like, there's a data scientist, there's like biz dev. And then I did like the pitch and we did ChatGPT and we made, basically we tried to prompt or we tried to embed smart contract vulnerabilities and fixes into ChatGPT so that you could just give it a GitHub link and then it would spit out your code commented with vulnerabilities and fixes. So, so it would fix it and tell you what it did in the comments. And, and so that was like our project, but I definitely see more and more like, how do we shortcut building? Like, how do we make building faster? How is this, there's this like middle layer? So when I saw Relay and I saw this like docs DevRel thing, I was like, oh yes. Like, so, so I just assume, and, or please just tell me yes and make me feel better. There's going to be like a DevRel version of this. And then I would think there would be like a customer-facing version of this. Yeah.

Yeah, there's, so customer onboarding, customer support, developer support, those are all each getting better. Yeah, basically there's Copilot, like GitHub Copilot.

Yeah.

Like that sort of future where, is it like, I like how you said the product people, like as a more product person, you now realize like, oh, you can basically have a team of devs. And I think developers are realizing they can now get a lot of help on the product side from this. And it's bringing the business together that way.

And also I think there's this need for, and this is what you're saying, I think with the chat is this ambient agent, having like an agent that's ambient. And I know that like, I have a friend, I have 2 friends who work at a customer service AI company. And this is what it does is they just have bots like running in the background, giving, like giving customer information as they're providing support or as they're doing sales. And so you kind of have always have this ambient assistant, almost like a heads-up display, but not, but in a way that you would naturally be able to interact with it. And designers are figuring that out. To where, to where you don't even have to, the idea is like not have like an abrupt, I would think frictional, like, oh, I need to ask AI moment. It should just be there. Like you said, with Copilot or any sort of assistant. Uh, although Grammarly gets a little, with that little green bubble, I have to move it around all the time because it gets

in

the way. Yeah, exactly. I want it to pop up when I want it. So anyway, there's like Raycast. I've been using that. And so, yeah. So Sean, do you have any suggestions? So you said like Tree of Thought prompt engineering paper, you talked about how there's like a data section, there's a prompt section and communication section of AI. Do you have any other suggested resources, events, things like for people who are interested in this AI crypto combo?

Yeah, so I think the 2 I'd recommend are, so the website for Relay is relay.cc. In the top header, there's a link to our Mirror, and the most recent post on our Mirror is Kind of a survey update of the blockchain AI landscape, which is about 20, 25 companies building in this space, hackathon projects. And a very interesting learning was, again, I think the post that spurred Bakelyst to get on AllianceDAO is, in my overview, I linked to theirs, which is their VC take of the intersection of AI and Web3. And I found that very illuminating. And I think those are the best 2 to start with.

Nice. I'm trying to mint it. And then I just noticed that you have like a layer 1 fee with Mirror and a layer 2 gas fee with Mirror. This is— anyway, so I posted those links in the chat and we'll have 'em in the— we'll try to get it in the podcast description, but this is awesome. This is really cool. Great, great job with this Mirror. Wow, it's beautiful. Cool, Sean. So we're gonna wrap up our time together, but I wanted to see if you had anything else you wanted to address About Relay, drop some alpha, things you wanted to promote?

Yeah, I appreciate it. I really appreciate you having me on here. Relay.cc has a number of bots. We trained those back in February, and so we're working on 2 major partnerships that I can't announce yet, but they're companies you've definitely heard of that will have more sophisticated chatbots. And we're also going to be launching, basically relaunching a sort of make your own bot for smaller startups to, one, have XMTP user-to-user chat on their site, kind of like a chat widget for that. And in addition, have a Relay chatbot that helps users onboard and does DevRel and that sort of thing. And so that's going to be coming out this summer. You can find us on Twitter @relay_eth and Really happy to be here.

So, Sean, if somebody wants to do a, like, talk to you about partnership, are they talking to you or is there somebody else they need to reach out to? Should they just hit up the Relay Twitter?

Yeah, that'd be me. My own personal Twitter is Sean W. Brenn. And yeah, I'm the founder of Relay. We have 2 co-founders and some developers and partnerships. And if you'd like to have your own fat pod built, I'd love to talk to you.

Awesome. Okay, so I will reach out to you about PoolTogether bot. We got to have it. Got to have it. Awesome.

Yeah. And it really just starts with documentation. That's, that's kind of the— people are finding out that's the hello world. That's for developers. That's like the kind of the easy thing to do with LLM. And so it's about one, productionizing that and making it scale to a number of users, a certain number of users. And the other side of it is adding a lot more features and functionality. Which I think we'll have a— I'll be releasing like a blog post very soon on Armir about kind of what I've talked about today, just how many ways one of these Web3-enabled chatbots can actually help a company or help a person. And so looking forward to showing that.

So, okay, before I let you go, so, so like, for example, with PoolTogether, we're releasing this new hyperstructure which will come out in July, and it uses ERC-4626 vaults. And so there's that, there's like the ERC-4626 Vault documentation. There's PoolTogether documentation. There's, we have James who works with District Labs and created Turbo ETH, and they're creating like a pool kit kind of thing. So there's that. And it'd be cool to like let all of that be embedded in a bot. I don't know what that is. Like if it's like a 4626 Alliance bot that has like anybody who's using for something like that. I don't know if that's like in your purview. I don't even know if purview is the right word to use there. If that's, if that's, but just, you know, like rather than thinking in like a vertical or like in a protocol, thinking as like a group or like, um, you know, like Lens. Lens has, I saw the Lens has its own thing, but there's, there's also a bunch of apps within Lens that are built on top of Lens. So I wonder if like I could ask the Lens bot a favor or an Orb question or a Butterfly, but yeah.

Yeah. Like once in, from what I've found in our testing, we actually do have some multi, like multi-doc bots that have multiple documentation sites into one. You do hit a limit where it starts to confuse Uniswap and SushiSwap and it's, it kind of melds the two. And so there is a way to get around that with some metadata filtering in the embeddings where based on the user query, it filters out some of the non-relevant Embeddings. But yeah, that's very possible to do.

Awesome. Okay. We're going to have to do that too. Well, Sean, thanks so much for being on the pod. It's, it's been a pleasure. Thank you for speaking to the PoolTogether community, sharing with us something that I'm very excited about, which is AI and crypto. So yeah, this is a historic moment. I'm going to look back on this in a year and be like, dang, I got to talk to Sean from Relay. So thanks for being here, man.

And when it's out of Billy, I'll be like, damn, I got to talk to Tim.

Yeah, that's right. Billy yearly. All right, cool, man.

Awesome. Take care, everyone.

All right. Cheers. Bye. Bye. Thanks for listening to the PoolTogether Community Podcast. You can visit pooltogether.com to deposit, and we'd love to hear what you thought about today's episode. So visit the PoolTogether Discord and let us know.