What’s Next for Social Appchains?

Social apps in Web3 have come a long way. Today, they feel much closer to the familiar Web2 user experience while still preserving the Web3 ethos of openness, ownership, and decentralization. Instead of being locked into closed networks where a single company owns your data and controls the algorithms, these new apps are built to be open-source and community-first.

The first wave showed up as NFT marketplaces with the likes of OpenSea, Rarible and others. They weren’t just places to trade tokens, they became social spaces where people could collect, show off, and connect over digital art. People started forming token-gated governance models(remember DAOs?). Social infrastructure projects like Lens took things further by building open social graphs and introduced a new window of opportunities in the web3 social landscape. Some other fronts, like finance integrations, identity and trust, and gaming, among others, also became major themes.

In this article, we’ll focus on Cartesi appchains and why they’re especially relevant for the future of Web3 social space. While many projects are pushing the boundaries of what’s possible, most are still limited by the constraints of today’s smart contract environments. Social applications, more than most others, thrive on complexity. Whether it’s managing identity, building reputation, curating content, or running creative, collaborative experiences. That’s exactly where Cartesi comes in.

Why Cartesi for Social Appchains?

Cartesi gives developers the freedom to build expressive logic that goes far beyond simple token transfers or fixed rules. Think about reputation systems that factor in identity, trust, and behaviour over time; recommendation and discovery engines that help people find the right content and communities; or even collaborative platforms where users can co-create stories, art, or games. With Cartesi’s Linux-based environment, you can build social gamified experiences or maybe run AI inferences(I built something on this, keep reading) to make social interactions richer and more engaging.

For builders, a huge advantage is the familiar developer stack. Instead of being limited to Solidity, Cartesi lets you use the programming languages, libraries, and tooling you already know — opening the door for a much wider pool of developers to experiment with social apps. I personally have built decentralised apps in Python and JavaScript. In theory, you can run anything that compiles down to RISC-V.

And the most important aspect, everything remains secure with a state-of-the-art, fraud-proof system, which runs permissionlessly and is backed by Ethereum. As mentioned in the previous article, Cartesi research contributors have released two widely known dispute resolution systems in the Optimistic rollups industry — Permissionless Refereed Tournaments(commonly referred to as PRT) and Dave. Heavy computations can be offloaded to the Cartesi Machine, making the computational burden and gas fees much cheaper without sacrificing trust. Needless to say, computations that are plugged into a fraud-proof system will gain more trust from the builders and the users alike.

Cartesi appchains also come with out-of-the-box modularity and integration with the underlying base layer. Developers can use the InputBox, Asset Portals, and Outputs to communicate with the EVM chains; switch between data availability layers like Espresso; and choose whether to deploy directly on Layer-1 or on Layer-2s, depending on their needs.

And of course, the best way to understand this is through examples. There are many experiments being done with the Cartesi stack. Some noteworthy social apps with asset-driven experiences are Comet, DrawingCanvas, or AI-blended projects like Scribbl. Others, such as RIVES or pkmn.fun, highlight the playful side of Cartesi-powered apps where games and social overlap naturally.

An AI-Judge Onchain - Scribbl with Cartesi 🤖

Scribbl is an innovative experiment that blends AI, blockchain and social interaction. The idea is simple: you draw a doodle, and an AI “judge” decides what it looks like. Your results are then ranked on a global leaderboard. It sounds fun and straightforward for the end user, but there is more to the core logic for developers here. Scribbl utilises the Cartesi infrastructure to make AI inference unbiased, decentralized, and fully verifiable on-chain. This means no central server can secretly influence the outcome; every judgment is reproducible and transparent.

The app shows how even a fun drawing contest can benefit from Cartesi’s infrastructure. By running inference inside the Cartesi Machine, Scribbl proves that verifiable AI can be part of social apps — ensuring trust while still delivering an addictive experience.

How it works:

1. Players draw doodles and submit them through the Scribbl frontend.
2. The images are registered on the base layer and read by the Cartesi backend, where they’re pre-processed and fed into a TensorFlow Lite model.
3. The model classifies the doodle, returning the top 3 predictions.
4. These predictions are ABI-encoded and submitted back to a customised Scribbl smart contract, which updates the user’s score and leaderboard ranking.

Scribbl takes advantage of Cartesi’s ability to run TensorFlow Lite models inside the Cartesi Machine. This allows the app to reuse open-source AI models while keeping computations verifiable and efficient. The model Scribbl uses comes from Google’s open-sourced Quick Draw dataset, which contains 50 million doodles across 345 categories, captured with metadata like timestamps, prompts, and player locations.

By running inference inside Cartesi, Scribbl demonstrates how social apps can use pre-trained AI models in a decentralized way, without needing specialized hardware or off-chain trust. The backend uses Pillow and NumPy Python libraries to pre-process user-submitted drawings, resizing them to a 28x28 resolution before feeding them into the model. The AI outputs its top 3 guesses, which are then ABI-encoded and passed on-chain.

Here’s a quick demo I recorded for Sribbl during a hackathon, check it out:

Looking ahead, Scribbl could evolve into a full-fledged appchain, where doodle challenges, asset logic (like NFT badges for winners), and the fraud-proof system all plug together to create a vibrant, decentralized ecosystem around a simple creative application.

Conclusion

I had a great time building Scribbl with Cartesi. It truly opens up an exciting new design space for social appchains. If you’re a builder in Web3, this is a new paradigm to experiment with the ideas in your head.

Reach out to me on X for any help. I’ll be happy to brainstorm ideas and get you started. Let’s go! 🚀