# MBD | Farcaster AI Personalized Feeds Source: https://www.mbd.xyz Kiwi News: https://news.kiwistand.com/stories/MBD-or-Farcaster-AI-Personalized-Feeds?index=0x6687e86b5b97e681e8f626224e1ddb474d4425fcd503c36cfbf8845d06d5ae1ae434f5c9 Submitted by yassinelanda.eth on 2024-07-05 UTC, 1 upvote, 0 comments ## Summary Embed is a product that searches, enriches, and ranks markets, tokens, trades, wallets, and social content for each user, then activates them through feeds, notifications, search, and agents. The page says it can go live in one to two weeks and requires no ML team. It describes a pipeline of Search, Features, and Rank APIs that turn wallet behavior into a personalized order, served through Feed and Notifications APIs. The page claims a 2.5x engagement lift from personalized discovery and that 50% of transactions come from personalized feeds, and it reports that its models were trained on onchain interactions. ## Article Embed searches, enriches, and ranks markets, tokens, trades, wallets, and social content for each user, then activates them through feeds, notifications, search, and agents. Live in 1–2 weeks. No ML team required. TRUSTED BY INDUSTRY LEADERS THE OPPORTUNITY Wallet intent is everywhere. Make it your moat. Without Embed — every wallet sees the same Embed turns wallet history into ranked product actions — before you lose the user. USE CASES Where wallet intelligence makes money. Activate new wallets Guide each new wallet to its first swap, bet, follow, deposit, or market. Cold start from 2 interactions Powered by Search + Rank + Feed Personalize discovery Replace generic feeds and static lists with wallet-specific ranking across markets, tokens, wallets, trades, and social content. 2.5x engagement lift Powered by Search + Features + Rank Reactivate dormant users Send the signal a wallet is most likely to return for, when it is most likely to act. Powered by Notifications + delivery guards Grow transaction volume Turn feeds, search, notifications, and market pages into revenue surfaces. 50% of transactions from personalized feeds Powered by Feed + Rank Retain high-value users Detect and serve relevant actions to traders, whales, creators, and high-intent wallets before they disappear. Powered by Features + Rank + Notifications Power internal ML Plug wallet intelligence into internal ranking, fraud, lifecycle, and growth models. Powered by Features + embeddings Latest From Our Team The team Built by a team that’s done this at scale Deep expertise in large-scale recommender systems, foundation models, and onchain data — across Web2 and Web3. Recommender systems at scale Personalization stacks that have served hundreds of millions of users. Transactions Foundation Models Event-sequence and behavioral models built for finance and crypto. Onchain infrastructure Production data and ML pipelines across major crypto protocols. The team has built and scaled at AWS Chainlink Gensyn Bloomberg JP Morgan Waymo " While LLMs predict the next word in a sentence, our recommendation models predict your user's next action. " onchain interactions trained on faster to relevant item (MRR@25) 0 domains: trades, bets, social MRR@25: How fast users find what they want Mean Reciprocal Rank: how quickly the first relevant item appears in the ranked list Backtest: last 10 interactions masked per user, model predicts next actions. No data leakage. BACKED BY DECENTRALIZED SOCIAL PIONEERS How it works One pipeline: Search, Features, and Rank turn wallet behavior into a personalized order — served through the Feed and Notifications APIs, all composed in the Algo Console. Onchain Chains data Offchain Search Generate candidates across markets, wallets, tokens & posts Search API Enrich Compute wallet-item features, similarity & preferences Features API Rank Rerank per wallet with trained models, then serve Rank API Your app PRODUCTS The building blocks. Six composable APIs. Use one on its own, or run the whole stack. Search API Find markets, wallets, tokens, and posts across crypto-native indices. Learn more → Features API Wallet-item signals, similarity, and embeddings for ranking. Learn more → Rank API Rerank per wallet with trained models, plus production controls. Learn more → Feed API Personalized feeds with configs, fallbacks, and caching. Learn more → Notifications API Signal-based reactivation with budgets, cooldowns, and webhooks. Learn more → Algo Console Build, preview, and deploy ranking — no data-science backlog. Learn more → Reliable & Scalable Performance DEVELOPERS Ship ranking experiments without waiting for the ML backlog Build with AI Describe the algorithm you want in plain English. The AI agent composes it — search, features, scoring, ranking — exploring your data with real user context. Or use the Embed Skill with Claude or Cursor to build locally. Get Started YOU Show me trending prediction markets ranked by volume, personalized to this wallet EMBED AI mbd.search() .index("polymarket-items") .size(50) .include([ { filter: "term", field: "active", value: true }, { filter: "numeric", field: "volume_24hr", operator: ">=", value: 10000 } ]) .execute(); ALGO BUILDER import { StudioConfig, StudioV1 } from "mbd-studio-sdk"; const mbd = new StudioV1({ config: new StudioConfig({ apiKey: process.env.MBD_API_KEY }) }); // Search trending prediction markets const candidates = await mbd.search() .index("polymarket-items") .size(50) .include([ { filter: "term", field: "active", value: true }, { filter: "numeric", field: "volume_24hr", operator: ">=", value: 10000 } ]) .execute(); // Score with ML model const scored = await mbd.scoring() .model("polymarket-rerank-v1") .userId("0xab2...e44") .itemIds(candidates.map(c => c._id)) .execute(); Build in the Console Compose algorithms visually in the Console — search, hydration, and DSL stages — or write DSL directly for full control. Preview results, validate, save your config, and get a Feed ID. Your feed is live. See Docs Serve in Production Once deployed, serve your feed with a single SDK call. Pass a wallet address, get back ML-ranked results at 250ms. See Docs FEED SDK import { getClient } from "@embed-ai/sdk"; const client = getClient( process.env.API_KEY_EMBED! ); const feed = await client.feed .byWalletAddress( "0xab2...e44", "your_feed_id", { top_k: 25 } ); Frequently asked. Got questions? We've got answers. Contact us at [email protected] for anything not covered below. What does Embed do? Who is Embed built for? Why use Embed instead of building in-house? How does the technology work? What can I build with Embed? How does Embed compare to general recommendation APIs? How hard is integration? Who's behind Embed? How does pricing work?