A Brief History of DOLR-AI and the Transition from YRAL to DOLR
We are three co-founders of YRAL, and our journey in the wild world of social media and blockchain kicked off in January 2021. Stuck at home due to the lockdown, we found ourselves mindlessly scrolling through hours and hours of content, especially on TikTok, which was the hot new way to get hooked on entertainment. That’s when it hit us—what if people could actually earn some kind of financial reward while scrolling through addictive social content? That single thought sent us down a rabbit hole. We started digging into research online and soon realized that while some people—about 25 million out of nearly 4 billion social media users—were making a living as professional content creators, and maybe 100 million were dabbling in content creation, the rest, a whopping 3.9 billion users, were just consumers, not creators. That stark divide lit a fire under us to start something big in the social space. Our mission? To make it so people sitting in the remotest corners of the world could earn at least a dollar just for getting entertained.
As we dug deeper, we found out that advertisers were raking in almost $125 billion a year, but all that cash was landing in the hands of a select few centralized giants like Facebook, TikTok, and Instagram. The profit from monetizing the attention of 3.9 billion users went straight to these primary stakeholders and shareholders, mostly parked in the U.S. That realization fueled our mission even more—we wanted to help social users earn even a single dollar, no matter where they were in the world. With the idea locked in, we had to figure out where to build it. We’d been small-time investors in crypto since 2017, so we knew blockchain was the only way to make a global platform with micro-payments work. We poked around Ethereum, all the Ethereum L2s, Cardano, and then stumbled onto the Internet Computer (IC) in May 2021, right when ICP launched and shot up to become the third biggest token on the market. Exploring the IC, we saw it was revolutionary—no other Layer 1 chain could touch its low-cost compute, storage capabilities, the ICO framework (SNS), and super-low transaction costs. It was perfect for our dream of micro-transactions with practically zero cost to users across the globe.
The Rough Start with HotOrNot
While digging into the IC’s tech, the only dev in our team @saikatdas0790, got hands-on with Motoko and built our first beta app—a single-canister, single-subnet dApp. Users could upload videos as challenges, others could join in by uploading their own takes, and content consumers could vote on which version was the best (say, 1st, 2nd, 3rd based on votes). Each vote was 1 HOT token staked, and the winners who guessed right would split all the staked tokens. With just a little marketing, we pulled in a few thousand users, but then the cracks showed. All the user data was crammed into one canister—an old-school setup—and the web app got painfully slow. Bounce rates shot up, and users couldn’t wrap their heads around the UX. We were just three people, with one developer and zero expertise in UI/UX, so it was a mess. Coming from humble beginnings with barely any resources, we knew we needed help. So, we applied for our first $25K grant from the IC, and when it got approved, we dove in with all the energy we could muster, building a better version of our social app one month at a time.
Since we had so little to work with, we picked a name that was catchy and didn’t need a ton of marketing—HotOrNot. User acquisition costs for consumer apps are brutal, and back then, we didn’t even have a token. From the first version, we also realized we had to make the game dead simple, something like TikTok, so HotOrNot was born. The HotOrNot trademark was up for grabs in India, so we applied for it and kept pushing our mission of letting content consumers earn rewards just for scrolling. This time, we ditched Motoko and switched to Rust because we wanted performance that could rival the fastest web apps out there. After some tinkering, Rust delivered, and the PWA started shaping up. The HotOrNot app mirrored TikTok’s UX, and we added a game where users could watch videos and vote if they were Hot or Not. A few months after launch, Dfinity hosted the Supernova hackathon (May 10 - June 20, 2022), and we showed off a content moderation feature in a separate feed where users could vote on whether a video should stay on the platform. See us at Supernova here (21:20). We didn’t rank, but it was a start.
Scaling Nightmares and a Viral Explosion
Even with the single-canister setup, the app started lagging as we added a few thousand users. So, we borrowed a page from OpenChat’s playbook, giving each user their own canister. That boosted performance big time, and scaling got a little less nightmarish. In December 2022, we rolled out a new feature—500 referral tokens for both a user and their invitee. That one move sent the app into overdrive. On December 6, 2022, we got 15,000 signups in just a few hours after some KOL from Indonesia dropped our referral link in a Telegram channel with over 100,000 users. The app blew up from there—tens of thousands of users playing the HotOrNot game millions of times. Dominic Williams even thought someone was DDoSing the ICP network! His tweet here. More on the madness here.
Early 2023 rolled around, and the SNS feature was ready—OpenChat was the first to test it out. HotOrNot was pulling in tens of thousands of users and burning through thousands of dollars in cycles, but we were scrambling to keep it alive with no extra resources. Dfinity invited us to try the SNS and raise funds to cover the insane cycle costs. We launched our SNS in May 2023, tested and implemented it by July 2023, and it was a massive success thanks to early Dfinity supporters. We raised 1.07 million ICP at around $4 each, giving us enough runway to keep building, grow, and provision cycles. Right before the SNS, we ran an airdrop campaign that nabbed us tens of thousands more users. By then, we had 300,000 users total—55,000 signed up—spending an average of 6 minutes on the app. Those 55,000 users meant 55,000 canisters on one subnet, and it started choking, tanking performance. We were also messing around with Rust-generated frontends via the Leptos framework, which beat out the React library we’d been using for frontend speed. Right after the SNS, we hit two huge roadblocks—scaling and frontend lag—and had to halt new signups while we figured it out.
Breaking Through the Pain to YRAL
The next few months were pure frustration and grit. We clawed our way to a single-user, single-canister, multiple-subnet architecture—the FIRST EVER on the IC. No one else had pulled this off. We wanted to kill the scaling problems that had haunted us since day one, once and for all. Alongside that, we built the first Rust-based frontend for a short video platform, making our web app as fast as YouTube Shorts (web version) or better. Our tiny team of five Rust developers worked day and night to make this happen—a web-based short video app that could hang with the biggest players out there. While wrestling with tech, we also had to tackle the business side, like legal headaches if HotOrNot went global. Over the last year, users kept uploading steamy videos, tied to the app’s name (we found out through user interviews), and the HotOrNot trademark wasn’t available in the U.S. Even though the U.S. wasn’t our focus, we knew the app’s viral potential could take us there. So, we put a proposal to the DAO, got community feedback, and changed the name to YRAL, hosted on yral.com (spelled Viral). Mid-2024, after a year of grinding through scaling and business challenges, we pushed YRAL through marketing channels, but hit another wall—onboarding users with Internet Identity was too slow for social media folks used to Instagram and TikTok’s slick signups.
That sent us down another rabbit hole to build our own homegrown identity solution. Our product team also suggested A/B testing to figure out which user cohorts responded best to app changes. Building those two features was months of pain and agony—nothing like it existed on the IC, so we had to do it all in-house. Over the last year, we grew from a five-member Rust team to 12-13, tackling every hurdle to make a scalable global social app on the IC. With A/B testing, we tried hosting the same app version on different domains to see what worked best—family-safe videos on YRAL, other stuff elsewhere. It was a huge hit. With our identity solution and A/B testing in place, users could share identities across assets with different frontends talking to the same canisters across multiple subnets—ANOTHER IC FIRST. Over the next few months, with more marketing, we racked up hundreds of thousands of videos on YRAL and expanded to over 500,000+ canisters across 20+ subnets. We were the biggest cycle burners in the ecosystem, blowing through $800K to $1M on cycles over the last year or two. We even built canister reclamation for inactive users—yet another first that took months because no one else had tried it. It was over two years of brutal hard work, pain, and agony to solve every issue and to be able to scale to millions.
The Leap to DOLR-AI
In early 2024, at an ICP Labs event in Zurich, someone floated the idea of an app for sharing user data between IC dApps so everyone could benefit from each other’s user acquisition. That stuck with us because we knew firsthand how insanely tough it was to build a scalable platform (years of effort, a big team, and heavy financial resources) and then grind for users through marketing. From that spark, and all the pain we’d endured, DOLR-AI was born. Announced here. With DOLR-AI, our mission isn’t just about rewarding users in remote corners anymore—it’s also about helping developers skip the hell we went through. The concept? Open our fleet of 500,000+ canisters across 20+ subnets to any app built on our backend. It’s still in the works, but soon, developers will just write frontends and hook into our canisters via APIs, tapping hundreds of thousands of users on day one. We’re abstracting the IC’s scaling challenges, making it 100X easier for anyone to build on the IC ecosystem. For users, picture an app store-like icon on every DOLR dApp—click it, share your data, and access other dApps without logging in again. It’s like a Layer 2 Scaling solution for ICP, solving scaling woes and giving new apps instant users. Vision details here.
This blew YRAL’s scope wide open. The DOLR token isn’t just YRAL’s utility token anymore—other dApps will use it to interact with the DOLR ecosystem and pull user data. YRAL will lead the pack, but explosive dApps on DOLR could skyrocket the whole ecosystem, pushing IC canisters from the current quantum of 900,000 to millions. Think of cycles on the IC—dApps on DOLR will pay DOLR tokens to run on our backbone, spreading cycle costs beyond just YRAL. Sharing users between dApps is a WORLD-FIRST, and we expect thousands of developers—building dating apps, health apps, games, you name it—to tap our 500,000+ canisters, which could grow exponentially with more dApps, not just YRAL. This will turbocharge the ICP ecosystem for everyone. We’ve already got multiple assets sharing identity on the same backend: yral.com (a decentralized TikTok), hotornot.wtf (AI-powered edgy content segregation), icpump.fun (instant token creation), and pumpdump.wtf (a P2P token game). We’re also eyeing partnerships with existing SNS projects to share users via simple APIs.
Wrapping It Up
All this came together after years of learning, hard work, pain, agony, and sheer stubborn persistence to build something massive—not just for us, but for the whole IC ecosystem. We’ve abstracted the IC in our vision so we can pull in users, investors, and developers from outside, growing the IC exponentially. Whitepaper here. With YRAL, we’re still chasing our original HotOrNot dream, but everything we’ve solved will make it 10-100X easier for other developers to build scalable apps on the IC. Our pain pays off for hundreds of others. We might not be the best at shouting about it, but our heart’s in making the IC a screaming success among other Layer 1s and we, as DOLR AI, envision us to be the biggest success case, both as a social app and a scaling solution on the Internet Computer Ecosystem.
At the end, our team really swears by the quote by Steve Jobs. “The people who are crazy enough to think they can change the world are the ones who do.”
Check out more on dolr.ai or hit us up with questions! We would be very happy to answer them!