How I Built Memescan with AI Agents (and Won My First Hackathon)



This content originally appeared on DEV Community and was authored by Ruben Marcus

The Idea

Last year, I had my “degen” phase — three months where I spent almost every day staring at charts, checking token lists, watching crypto YouTubers, following KOLs, and joining Telegram communities with people all over the world.

It was stressful, but exciting.

The speed of memecoins is insane. News happens anywhere in the world and, before it hits a major outlet, a token is already live on Pumpfun.

After trading and following these communities for a while, I realized it wasn’t for me. Still, as someone who loves internet culture, memes, and crypto in general, I kept watching the space daily, even if I wasn’t trading anymore.

But one thing kept bugging me: good tools for traders were missing. The space feels rigged. KOLs, whales, and scammers have all the advantages, while the average trader with $10–100 in SOL is left powerless.

That’s when I started thinking: what if I built something that leveled the playing field?

The Challenge

Last month, my company — Bitte Protocol — hosted an internal hackathon focused on AI Agents. The goal was to explore what AI Agents could do in blockchain, leverage our user base, and test adoption.

I already had the idea of building a memecoin tool because my friends (who still trade daily) constantly complain about how bad current tools are, and how easy it is to lose.

At the same time, I was experimenting with new AI Agent ideas every week. I felt that using AI only as an assistant wasn’t enough — it needed to integrate naturally into web2 flows. People shouldn’t need to learn web3 just to benefit from it. It should be optional, not mandatory.

What is Memescan?

That’s how Memescan was born.

The name plays on the word “scam” — on purpose. It’s both a joke and a statement about the scammers that plague crypto.

My goal wasn’t to build the perfect, revolutionary tool in a week. Instead, I wanted to make something better than what was already out there, for a community that clearly wanted it.

Imagine if users could:

  • filter tokens,
  • vote on whether a token is worth aping into,
  • highlight strong communities,
  • and actually have metrics to make better decisions.

That’s Memescan.

We also have a Telegram group where the community makes calls on tokens they think will boom.

Why It Matters

Memecoins are the new internet money. For people born after the 90s (especially 2000s), growing up with mobile phones and the internet, it feels more natural to speculate on tokenized assets than to walk into a bank and ask for a savings plan with terrible returns.

Memecoins themselves are not “investments,” but for a younger digital generation, investing in the culture they live in — memes, internet jokes, communities — makes more sense than investing in the S&P 500.

If you’re new to memecoins, this video from Murad at Token 2049 is a good start. He’s extremely bullish (and biased, with a $50M+ portfolio), but he makes some interesting points.

What I Think About Memecoins

To me, memecoins are:

  • Pure gambling. The odds are 9-to-1 against you.
  • The most volatile asset that ever existed.
  • Community-driven. The real fun is the collective energy people put into it.
  • An easy way to tokenize any idea. You can launch without upfront money using a launchpad and earn from fees or creator payments.

This power can be used for both good and bad. Yes, memecoins are often linked to scams, stupidity, or even crimes. But there are also builders and communities genuinely trying to create something fun and cultural.

Even if 99.9% are in it just to make money, the 0.1% are building. And that’s what inspires me.

How I Built It

I had already experimented with memecoin agents on Bitte, on Near Protocol. This time, I wanted to build on Solana because the user base is much larger.

For Memescan, I started with:

I stripped away what didn’t fit, and added what I needed.

The Agent Side

I built the basic feature: scanning tokens.

  • Gather info from Solana RPCs, launchpads, Dexes, and APIs
  • Run calculations to show volume, market cap, pricing, risk analysis, etc.

That part was relatively easy.

The Hard Part: Frontend

Splitting the chat package between two screens was tricky.

I realized the chat embed (text input + chat messages) could work as:

  • Home page → search box
  • Token page → analysis generated by the agent

Why force everything into a “chat” UI when I could turn it into a search/result interface?

We can still keep the AI Terminal setup, but it doesn’t have to be the only way users interact.

Token Calls Feature

I also added a “token calls” feature (not yet AI-powered, still web2). Building and maintaining it was tough, but it’s a core piece of the community side.

What is Bitte Protocol?

Bitte Protocol is the company I’ve worked at for the past 3 years. It’s also how I entered crypto.

We’ve been building a stack that lets developers (or even non-devs) create agents for blockchain.

We believe:

  • The agentic economy is already here — it’s only going to grow.
  • Blockchain and decentralization are crucial to that vision.

If you’re curious, check us out:

How It’s Going

Honestly? It feels amazing.

I finally have a tool that pulls together the data I always wanted, stores it, and feeds it into my AI Agent.

But it’s also exhausting. I’m handling everything: marketing, development, AI testing, bug fixing, outreach on Telegram and Twitter. It’s tough but rewarding.

So far:

  • 5,000+ scans
  • 1,020+ calls
  • 30k+ AI Agent pings (in just 2 weeks!)
  • Users from 40+ countries
  • Indexed on search engines, launchpads, and communities
  • Daily active users: 50–150
  • Some community calls have already hit 100x gains

Of course, scammers are also launching fake tokens using my app’s name — but I expected that.

The best part? People DM me saying they use Memescan every day. That’s priceless.

What’s Next?

Plenty:

  • Trading features
  • AI bots
  • More advanced AI analysis
  • Partnerships with communities

It’s hard building something serious in a space known for scams and rug pulls. But that makes me even more determined to succeed.

Want to Know More?

I’m just getting started — Memescan is a bet on culture, community, and technology. If you care about that too, come build with us.


This content originally appeared on DEV Community and was authored by Ruben Marcus