Origins Of Memescale

My name is Chris and I am the creator of Memescale AI. For me Memescale was a self fulfilling prophecy, that took over a year to fully build out the way I think people could enjoy. To reach me directly, message @ChrisScaled on Telegram, or @CheifChad on X. For bugs or feature requests, it's best to use those dedicated channels. For inquiries or other questions my DMs are always open.

The year is 2024 and I am seriously starting to take memecoins seriously with the rise of pump.fun, the daily volume started growing exponentially onchain and it was clear there was money to be made outside of strictly holding majors which I previously hadn't believed. This led me to research Solana specifically, the biggest coins, and learn about how they got big. I looked into what made coins big because I knew I wouldn't enjoy trading memes as I prefer believing in something powerful that brings people together, differences alike, where everyone is putting their energy toward the same movement with one shared goal. After getting familiar with Solana, I slowly started to realize it was about investing in the right people, and coins, rather than only the right people or only the right meme. In some cases, it certainly is only about the meme, and in other cases it certainly is only about who is behind the meme, but I wanted to invest in both.

This is when I stumbled across a project called UberJeets, and thought the meme was absolutely hilarious. For context, Jeet is not a racist term in the way it is commonly referred to in the mainstream and has nothing to do with race. A Jeet is anybody that sells a large amount of tokens usually for a short term gain, while others are still holding the bag and believing. Jeeting isn't bad, it's how people make money, but the joke behind it is bringing humor to that horrible feeling of, "what if I would've held", and this coin UberJeets was hilarious. An original meme with a crying delivery driver that Jeeted working for UberJeets, and this was one that I bet on. Initially, I was a community member that knew nobody, I wasn't in any private groups, wasn't following anyones twitter to tail a call, I was simply On DexTools scrolling through coins that were a few days old but for some reason have not died yet. As time went on, I started talking to the community, the CTO leader, and realized what we were building was more than a meme it was a community that believed in Solana, a community that believed in each other, and most importantly a community that believed with all of these elements combined could make it happen.

As soon as I got more familiar with the community I wanted to do more to help grow the project. But I had to face it for what it was, I had zero audience, zero followers on twitter, zero reputation, so I needed to make a change. While I was figuring that out, the community continued growing, we hired a developer to create the "Jeet Receipt" Solana utility that helps you calculate the exact amount of money your wallet would've made had you not Jeeted that token. UberJeets was thriving and making real progress as a grassroots CTO. As time went on, the communities commitment grew deeper and deeper to the point where people were starting to notice us, we had hit an all time high of 1.6m market cap, entirely organically, with zero paid promotion just by spreading our energy and mission across social media. At this point, UberJeets was undeniable in my mind, my bag had grown from what I would be comfortable losing to an amount where I realized my earlier thesis about memes being worth it to hold alongside majors was validated. So I continued believing. I learned how to shill on Twitter spaces. I onboarded friends, old classmates, and even my friends girlfriend to this project.

As UberJeets evolved, we realized the IP of this meme was unique and wanted to introduce a new character, we called it the Chad character. This chad was the version of the Jeet that never actually sold his bag and because of that he saw massive success. This was the cool character people aspired to be, the complete opposite of the Jeeter that sold his bag. This character was instantly popular to our community and gave our project a unique storyline, something we could make content with, and even short form animated videos showing how the Chad and Jeet both came to be. From a random community member that was working for a project I found on the internet, to funding a press release about our vision for the coin, I knew I wanted to continue pushing this project, but everything started to get expensive. The music started to stop, the $10 for every meme started to add up, new members stopped joining as frequently.

But what were we doing wrong? Not knowing certainly didn't help, so we doubled down. We pushed harder, we stayed on spaces longer, we scheduled more voice calls, we made more plans, we found more connections, and we realized that this requires sustaining attention on something people naturally don't care about for long. That was a difficult thought to process, but it didn't even faze us. We needed revenue. We built a business. We hired a volume bot developer named Impulse and planned to build the most affordable volume bot on Solana. We saw the numbers that fatality, orbit, and others were doing and realized Solana is only going to continue growing, there is room for us in this market. So we designed the entire idea, we talked about it everyday, we made clear instructions for the developer, but the development updates kept getting more and more scarce. Instead of hearing from the dev every other day, we heard from him twice a week, then once a week, and so on. Every single date we had heard for testing was missed, until finally I realized that when he eventually does have the backend ready I can build a frontend now so we can instantly go live, the community was fed up as it was and done with waiting. So in the midst of the developers absence, I built a decent frontend for this volume bot just in hopes that maybe he'd come back, deliver, and the frontend would be waiting for him.

That didn't happen, and never did. To this day I have never heard from that man. After ghosting us for long enough, his Telegram status turned to last seen a long time ago, and I realized that shit was absolutely cooked. It hurt extremely bad getting betrayed like that, after he was working for 11 months. We had sent him large amounts of money, bonuses, I had even sent him almost a % of the supply on a random Sunday just in hopes that'd motivate him to continue working hard. It didn't. I did not understand how that was possible. Everyone is motivated by money. So after a few days passed of that last seen a long time ago status I realized the entire business idea that relied on this volume bot was now dead.

With a community getting more and more frustrated, I didn't want to just tell them it'd be fine, because it wasn't.

I needed to do something to actually let the community interact with, not only to show them that we're truly committed and building, but to grow our community into something that was fun to stay in, not one that has members saying "GM" back and forth so new people join and don't think it's dead. So I starting looking for things to give the community, in any way I could, and finally found it, and it was absolutely perfect. An AI Meme Bot, a bot that lives inside of our telegram group, that generates images of both the Chad character, and the Jeet character for anybody to use. I started getting Chat GPT to generate images of the chad, I came up with a two paragraph prompt that would get it to output something that came really close, but after re-evaluating I realized it really didn't come that close, this is an entirely different character. The first option I found was a decently reputable Company that I had previously met with on video calls, and found to be trustworthy, so I reached out and asked how much something like this would be. Not only did their answer knock my estimates out of the water, but there was real weight attached. This company was going to charge me 10 cents for every single image the community generated in the form of an image credit in addition to a $750 setup fee on the bot. Worst of all they required me to purchase a significant amount of image credits upfront to start, and it was not practical for the project at the time. Next company I looked at, $20,000, for the entire setup, a custom bot inside my group generating characters of the image. Was about to ask if that included the Blackwell running it, but held off on that. I was starting to understand that there was nothing for little fish.

I began looking into getting my own meme bot built, with an API attached to the latest nano banana model, and a good reference image of the Chad and Jeet. The first image that it generated was like magic to me. I had never worked with image model APIs and the Telegram bots I had experience with weren't sending and receiving requests the way this was, it made me feel alive. The first few days of the bot, it was delivering image quality I had never seen before, they were almost as good as what I had paid our illustrator for at $10 a pop. But it just didn't feel right. Every morning I woke up the first thing I checked was Telegram to make sure nobody was abusing my bot, costing me money on the API. Refreshing the balance of the API became something normal since this instantly became popular in the group. Members that had been in the group for almost two years at that point but never said a word started prompting this bot. And then it really hit me. This group has like 70-80% of its members inactive, but lurking on occasion, just waiting for anything that'll interest them, or be of importance to them, and for some this really is close to nothing. However, this showed me something that opened my eyes.

Then one night I woke up and someone had generated 700 images with my bot, no admin online to stop it, no mods banning them, just pure malicious intent. I started realizing that this isn't sustainable. For a small CTO that isn't profitable, or selling any tokens at all, to continue pouring money into a project for years on end became difficult. So I really didn't know what to do. It felt like the entire bond we had built in 2024 and early 2025 started melting away, and was hanging on for dear life by a string. Chat GPT wasn't capable of holding the IP of the character on every image, every choice I had was minimum thousands with weight and image credits attached, even getting quoted $20,000 by one company. As the project started to slow down in mid 2025, that's when I truly realized this is up to me. Nobody is coming to save us.

So I started learning about how image models work. Started learning what latent space is, what pixel space is, what a text encoder is, how natural language gets vectorized, how the denoising phase works, how to leverage KV cache with minimal loss, and how to maximize speed without sacrificing quality. As Toly says himself, increase bandwidth, lower latency, that was the entire idea. The journey felt like a roller coaster, in the beginning I felt like this would be extremely easy if I study hard and listen to the right guides. Then night after night, week after week, to month after month I was humbled. Humbled about what it took to build a meme bot, humbled about session management, humbled about systems provisioning, humbled about using the local telegram bot api, humbled about what it would take to even make that first LoRA. While putting my last breath into building this, UberJeets had died. Not because we didn't want it bad enough, but because the market didn't care how much we did. I was attempting to build this bot for the one project I had been a part of throughout the entire bull run, and as a result I was thrown in the dirt, not by the community, but by my own ignorance.

I failed at making UberJeets successful and I can acknowledge that. Everything I said UberJeets would be, to all of my friends, to people I had onboarded didn't come true. I started refunding almost everyone I had personally onboarded to the point where the bull run didn't make me a dollar, and in fact had a fairly devastating effect. So I was pretty sad for a while, felt like everything I had been pouring my energy into was thrown into a fire. Every space I had ever shilled on seemed like wasted time, every post I ever made seemed like a pathetic shout into an echo chamber that was designed to reject pathetic things, and worse of all I had lost a shit ton of people money, and was fully responsible for that. I didn't want to stop building. I wanted to finish what I started, unlike the volume bot developer that threw us in the dust after 11 months. I saw how our group transformed those few days I had a paid API attached to the bot, I saw new members joining out of nowhere to use it, and I saw inactive members come back from the dead. This is now Q4 of 2025, and I had a basic understanding of diffusion, knew how to train a model, but needed to dive deeper into how the best outputs are created. I named my project memescale, because I wanted to build a company that was truly designed to do nothing other than scale memecoins. So I started going all in. I learned exactly how to create deployable on-demand servers that could power an entire army of bots that can be edited, updated, and deployed within the click of a button. I looked to my left, and then looked to my right, and didn't see anyone that could possibly stop me. The only one that really had that power was myself. After it had been nearly 7 months of studying, building, and researching it was finally 2026 and I knew memescale could transform a lot of peoples Telegram groups. However, it felt like after every big step forward there were two steps back, and the steps back kept hurting more and more. Memescale wasn't just a bot to me, I wanted to build a platform where people can instantly deploy bots, and then manage them in real time. Add commands just like other bots, but with more features. I wanted to make rate limiting and banning users from the group as easy as pushing a button, without even needing to kick people out of the telegram groups and I did. I wanted to make it so that when you update the profile picture on your bot in Telegram, it updates instantaneously on our website without even refreshing the website, and I did. I wanted to make session revocation instant as well, because when you revoke a session, it needs to be revoked then and there, not just get rejected on their next API request, and I did that. I wanted to make it so that when you train a model on a meme bot, there isn't a waiting time where we analyze your images.

I wanted the model training to start that second you click purchase, so that if you purchase a bot at 2am, its ready by 5:30am latest, and did.

I wanted to make it so that technical users can bring the API token of a bot they already own too, not just have us create the bot for them, and did.

I wanted to go above and beyond for anyone that would ever use Memescale and that was the only thing I cared about.

I was putting more time into this than I was at anything else in my life. The only thing I could think about was how can this be better, how can this be more secure, how can I exploit this from a different angle, how can I attempt to hijack my nginx with spoofed headers, how can I spoof the hmac to accept a request on a bot change from a random browser, and was never satisfied. Nothing was impressive anymore, no feature was enough, and that's how it is until this day, You don't know you know nothing, until everything you don't know becomes clear, and then it becomes humbling. I studied more, practiced more, built more, deployed more bots, trained more models nobody asked for, generated millions of images, spent time learning how to exploit weaknesses in cryptographic session security, and learned how to hack forge cookies and brute-force hmac validation. Brute forcing and hacking taught me that cryptographic proofs for session security and endpoints are only as good as the re-validation of the original challenge upstream. All this, just to make my own system stronger. Perfect, no, but stronger, yes. I wanted to build something that was timeless, not something that the next Claude model will be able to find a zero day exploit on. Coupling this with what I already learned about Telegram systems provisioning, I realized that I can deliver real value that can help memecoins and other projects without having to worry about their bot getting hacked.

Then, fast forward to mid Q1 2026 session security was finished, and confidently so. Then it became time to put my all into the only thing that mattered from there on out: image fidelity, speed, and quality. This is where the humblings would go from "painful" to excruciating, because training models is not something you learn once and click forever. Every base model is different. Even one string on the config can drastically change how a model looks, aside from the dataset. So I trained hundreds of models, of characters from tons of different projects, and began developing a training system that I was confident in. Now, I am training decent models even on horrible datasets, which is why memescale immediately trains your model when you purchase a bot. Memescale uses a deterministic approach that doesn't have any AI models telling it what to do, theres no chat bot saying load up the model, everything is sequential. From the moment you pay for a bot, to the moment you generate an image of your character in the group, there isn't any intervention from myself. If I tragically passed away tomorrow, memescale would still be live to train, host, and operate as a platform you can rely on day in day out.

While downtime is extremely rare on our end, programmatic safeguards automatically execute 30 seconds after a GPU first goes online, and provisions a fresh machine where your weights (custom model) are transferred, and then live within roughly 10-20 minutes depending on network speed for the model download. While I am now focusing on free Telegram tools, and more features, I know there are projects out there that will benefit from having a memescale bot in their group, whether that be to unite their community, get old inactive members active again, or to simply generate hundreds of thousands of memes a month for $99. Memescale has an extremely rich set of tools and functions that go far beyond image generation, and image editing. Telegram assets that everyone knows about, like those little 3d emojis that have the spinning coins people put by their name, or those little banners that use animated emojis to form words, animations, or other menu-like usecases, used to require a very talented animator, hours of work in some cases, money, a real blender-like environment, and they are now entirely free, for all. Regardless of if you have a memescale bot or not.

You can even use this on your free memescale bot that we offer, by sending /newpack, or using our website, you too can create those coins that people are still to this day charging $10-$20 for each.

For the menu-like animated banners you'll sometimes see on telegram, those are comprised of premium telegram emojis that when put together can host very cool animations. Each emoji can have up to 64kb on it, without hacks to get bigger files hosted, which is enough for complex animations all of which are offered on our website. On the website, you'll see a "Free Tools" button that opens a menu that shows the coin builder, and banner builder, each of these will allow you to fully customize your asset in a lightweight environment that even low-spec users will be able to get decent performance on. After you've finished, depending on if you have an existing telegram bot with us, even our free bot, you can choose to export it by copying the emoji pack link which you can paste right into Telegram, or you can have your bot send you the new emoji straight to your DMs. For technical users, developers, vibe coders, all in between, you can also export the lottie json which is helpful for embedding on mobile apps, loading screens, websites, and more because they're under 64kb. Memescale has developed a decent formula for vectorizing the image which allows even the most complex images to retain a lot of the detail once it's in lottie. Throughout internal testing, we first noticed that complex images were getting extremely distorted, and simple images were creating great coin emojis, so days turned into weeks building something that would create nearly lossless vectors on most all images. You can try this out right now.

This is the end of the story. Memescale is a random project that barely looks legit, so I wrote this explaining every part of my journey for those who are skeptical of the "unlimited images" nature and pricing dynamic. Unlimited just means we have GPUs running for you that continuously generate images without ever stopping, your plan is renting reserved capacity and nobody can jump ahead of you. On the $99/mo plan, you are sharing a GPU with 3 other projects, and queue is round robin, so even if one project has 100 images in the queue before you've sent a single prompt, and you generate 5, you guys will go back and forth until your 5 are generated. This means at the bare worst, our $99/mo plan is generating images for your project every 30 seconds, with no limits. For anyone that has made it this far, I appreciate you reading this story and checking out memescale. I highly encourage you to use our free tools, mess around with them, provision a free bot, edit some images, check out the presets, use it as a group moderator that automatically cleans itself up and doesn't leave messy messages in your group (This is off by default unless toggled on) or simply use it to send Youtube, Soundcloud, or Tiktok videos into your group or to the bot in the DMs where it'll send you the audio file back to you, with customized metadata so that your bots profile picture is the image that shows on that file.

All of this is free, and will always be. The $99/mo plan simply introduces character training, and the unmetered aspect. Free bots can edit 50 images per day and only work inside of the groups you specifically authorize - just like our meme bots. Because random people shouldn't be able to hog your queue of your bot, only you and your authorized group should be able to do that. Additionally, all of our paid plans come with even more free stuff too, most of it not even advertised. Because we didn't build memescale to impress anybody or to take away from competitors, we built it to support real communities in the best way possible, because if I could go back in time two years ago to when the volume bot developer wasn't hitting me back, I would've wanted the exact same thing. I was never able to get that, but you can. Thank you for reading this.