The Investment

We met Yassine from ❜mbd in August 2023. Back then, Yassine was building an ML Models to create a feed of personalized Mirror, Farcaster, and Lens posts with the vision to create the web3 social recommendation layer powered by machine learning. He was part of the Chainlink AI/ML team and previously built ML recommendation systems at AWS, we knew he had the technical and ML and Web3 chops to make this a reality.

After graduating a16z’s CSX23, we decided to invest.

In 2023, content moderation was a $9bn business. A majority of this is for the “Trust and Safety” sector. The large majority of recommendation engines are proprietary to the incumbents and are in a black-box (so hard to estimate market) - but the market size is estimated to be $30bn+.

❜mbd tackles these giant markets for web3. We are betting on market expansion of web3 social. We thought the initial model of digesting all available web3 social information for builders to personalize feeds was great, but we were mostly impressed with the larger vision of a protocol where protocol stakeholders would be able to provide and label data that would be of benefit to others.

The three use cases we saw were:

1. Content moderation: Trust and Safety teams of web2 do this algorithmically, but also manually, we think a decentralized model would be much more efficient for labeling these types of content.

2. Recommendations: All the way from posts, to LLM chatbots, all the data helps to customize these.

3. Community notes and fact checking: It’s essential to fight our post truth era.

Today, ❜mbd serves 100+ teams, from larger to smaller teams who are in the search of easy to use APIs for onchain recommendations. Teams have also built +300 feeds on the platform and can manage the user experience from a no-code interface that will soon be available to consumers. Just in their beta phase they did 500k+ recommendations. 

❜mbd is the category leader by a landslide, and built a scalable and perfomant ML platform for moderation and personalization that can deliver 2x the engagement and transaction volume and a tech stack that can do billions of requests under 150ms response times. This was cemented early 2025 when the Coinbase Wallet team announcing they are integrating a social feed into the wallet powered by ❜mbd, launching soon this year.

Company Building

Here’s a conversation with Yass.

Q: Why did you decide to build on the AI/LLM space before it was ‘cool’? How has it paid off to date?

A: I’ve been fascinated by programming since childhood, so when I was in engineering school (2012–2016), I saw the AI/ML wave coming. Neural networks were just starting to scale after decades of AI’s “dark winter,” and I knew I wanted to build in this space.

In 2018, I launched my first AI startup—building music recommendations powered by AI that could understand your private context through biometrics like heart rate, temperature, and even brainwave signals. The AI worked, and it was clear it would only get better. But privacy? That part was still unsolved and wildly underinvested in.

That gap led me deep into cryptography and blockchain. Fast forward to today—crypto infrastructure has caught up, AI has exploded in power, and combining the two is a no-brainer. Investing years into both fields is deeply embedded in ❜mbd’s DNA, and it’s what will keep us ahead for years to come.

Q: What is your current sales / user acquisition motion? How did it evolve over time?

A:  Early on, we knew we were tackling both technical and market risk. As a team of veteran data scientists, we were confident in solving the technical challenges, but the market was still nascent. So, we focused on getting as much real-world feedback as possible—building alongside the bleeding edge of native crypto/social users.

We started by being our own first users, participating in hackathons to dogfood the initial alpha. That hands-on approach helped us deeply understand developer pain points. Then, we leveled up—hosting our own AI and Web3 Social hackathons, running in-person events with the beta product.

At EthCC 2024, our ❜mbd beta APIs powered projects that dominated the AI hackathon—every builder using them landed in the top 3 finalists across all tracks (even ones we didn’t sponsor), while only 30% of the product was functional. That was the validation we needed.

With growing confidence in the stack, we moved upmarket—testing demand with large crypto consumer apps, locking in NDAs with all major wallets, and securing Coinbase Wallet as a design partner for their new app launch.

Q: What’s next for ❜mbd?

A: Better AI models, big moves in Agents (our fastest-growing customer segment), and a surge in crypto content to power the next generation of social apps.

We just launched onchain content feeds on Base at ETHDenver 2025, and we’re bringing that same ML-powered personalization to NFTs and Tokens next. More social data is coming soon—from Web3 networks like Lens, new chains like Solana, and even mainstream open networks like Bluesky.

We know exactly what needs to happen to make ❜mbd the go-to personalization platform for Web3 and Web2. And we’re building at full speed. 🚀

This post is for informational purposes only, and does not constitute a recommendation to buy or sell securities or to pursue any particular investment strategy. This post should not be relied upon in evaluating the merits of any investment or any particular investment strategy. You should consult your own advisers as to business, financial, tax, legal, and all other related matters concerning any investment. The views expressed in this post reflect the current opinions of the authors and do not necessarily represent the opinions of Social Graph Ventures LLC