❜mbd raises $3M to bring AI-powered recommender systems to crypto consumer apps

We’re excited to announce that ❜mbd has raised a strategic pre-seed round of $3M co-led by Mask Network and Polymorphic Capital with participation from a16z crypto CSX, Forward Research, Social Graph Ventures, and WAGMI. ❜mbd has also integrated Amazon Web Services and Google Cloud AI Startup Programs.

We’re proud to have such pioneering teams, who champion the open and decentralized social movement, and believe in our mission to build the intelligence layer for content moderation and recommendation.

Recommender systems: the missing piece for engaging Web3 social apps

Decentralized social networks like Farcaster, Zora, and Lens allow builders to access extensive data from day one by building permissionlessly on an open social graph. This eliminates the cold start problem that typically arises when building social applications.

However, creating engaging Web3 apps—whether wallets, NFT platforms, or AI agents—requires more than just data access. Social apps must identify and deliver high-quality, personalized content within their vertical to engage users from the start and sustain their interest. They also need scalable and reactive moderation systems to fight against spammers and other bad actors.

While rule-based systems and manual curation may help, an efficient content engine ultimately depends on advanced,machine-learning recommendation systems like those we specialize in at ❜mbd, designed to deliver dynamic, relevant content and actions tailored to each user.

Unlocking blazing-fast, high-quality, and hyper-personalized content for Web3 social

At ❜mbd we’re building towards a decentralized moderation and recommendation AI protocol for low-latency consumer applications. Our flagship product is a set of machine-learning APIs for Web3 developers, designed to deliver relevant, timely social content, mints, and transactions to users by incorporating both onchain and offchain data.

Using custom machine-learning models that update their memory in real-time, we offer an advanced recommender system that boosts user engagement and retention, providing Web3 developers with feature parity comparable to web2 giants like X/Twitter, TikTok, and Facebook.

Our product is accessible now through a no-code feed builder and API, allowing anyone to easily create personalized feeds filtered by dApps, channels, topics, authors, and more. Since our private beta launch in late 2023, focused on Farcaster content moderation and recommendations, we have onboarded hundreds of developers. API usage has grown quickly, currently sitting at 100k+ monthly requests on our public beta.

The custom AI models we’ve built are trained on onchain data from Farcaster, Lens, Zora, Base, Ethereum, and Arweave. Our latest generation of ranking models, with over +100k hours of training, can predict what onchain users want in real-time, boosting relevancy by up to 3x and content coverage shown to users by 10x compared to popular feeds.

Finally, our classification models for content moderation effectively remove spam, farmers, and other malicious actors, achieving performance on par with leading AI models from companies like OpenAI—at a fraction of the cost—by leveraging Web3 fine-tuned open-source models.

We believe ❜mbd is the most advanced AI stack in production and available to web3 developers small and large. With real-time personalization that updates model states every time an onchain interaction happens keeping freshness high, +50 categories scored in real-time as content gets created to provide powerful curation and moderation filters, and an AI inference layer that serves recommendations in inder 150ms at the scale of billions of requests, we are ready to have feature parity with the best web2 apps.

Our next steps to support the future of Web3 social

As on-chain data expands exponentially, we see a unique opportunity to catalyze the explosion in the number of social apps built on top of open social graphs. Our commitment is to build the most modular, powerful, and easy-to-use intelligence layer for on-chain content, boosting client diversity and bringing crypto consumer apps to the masses.

Our immediate focus is partnering with crypto consumer applications to bring actionable and fun onchain content to users. We’re starting with Farcaster content and progressively expanding our model coverage to more onchain users from different ecosystems. In the long term, we’re dedicated to shaping the foundations of Web3 social and addressing some of the biggest flaws of web2 platforms. We have already made great strides in unlocking permissionless community-driven moderation and verifiable recommendation systems through extensive research. Our goal is to pioneer a new design space for social media algorithms, which we refer to as “social epistemology”. We are excited to be sharing more of our findings soon.

To achieve our goals, we’re using this round to expand our team with key hires who bring deep experience in world-class mobile experiences, disruptive go-to-market strategies, scalable AI systems, and decentralized protocols. We’re proud to have a team who worked across both Web2 and Web3 at leading companies like Amazon Web Services, Samsung, Chainlink Labs, and Gensyn. We’re always eager to connect with mission-driven engineers, researchers, and builders passionate about Web3 social and decentralized AI—if that’s you, we’d love to chat.

Thanks for being part of this adventure! If you are a Web3 social builder looking to bring engaging onchain content to your users you should visit the links below or reach out to us in DM on Farcaster or X.

Links: