{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/6cf7c88f009a48e5b08d590f4de6e9b3\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/6cf7c88f009a48e5b08d590f4de6e9b3-dbf0ebbf80ecd130.gif","duration":4365.567,"title":"Office Hour #6 – Hyper-Personalized Leads Engagement, Telegram Bot How-To &amp; MCP 2.0 🚀","description":"In this jam-packed Office Hour #6, Sean (Founder of AgenticFlow) walks you through some of the most powerful updates we’ve rolled out:\n\n🔹 New Knowledge Engine Revamp – with a top-ranked embedding model (from Rank #166 → #5!)\n🔹 Enterprise-Level Search & Re-Ranking – for faster, smarter hybrid search workflows\n🔹 MCP Goes Mainstream – ChatGPT, Cloudflare & Vercel support, and how you can build your own\n🔹 Hyper-Personalized Lead Engagement – with LinkedIn scraping, canvas-based image gen, and Gmail draft automation\n🔹 Telegram Bot Tutorial – complete walkthrough of how to create and connect your own AI-powered Telegram bot\n🔹 Live Q&A – covering multi-agent updates, performance tips, and roadmap insights\n\n📌 Plus: Sneak peek into multi-agent beta rollout (launching July 2025!) and an exciting plan to turn AgenticFlow into its own MCP server—so ChatGPT can run your workflows 🤯\n\n📝 YouTube Description (with clickable chapters)\n0:00 – Warm-up & Audio Check\n1:07 – Today’s Agenda (what we’ll cover and why)\n2:26 – Knowledge Engine Revamp – new Link-Q 4000-D embeddings (Rank #5) and why your relevance score should drop from 0.5 → 0.2\n3:53 – Enterprise Re-Rank + Hybrid Search – cut tokens, speed queries, boost accuracy\n5:52 – New MCP Connection UI – SSE vs Streamable HTTP and easier header handling\n8:08 – MCP Goes Mainstream – ChatGPT beta, Cloudflare Workers + Vercel templates, 2K+ first-party endpoints\n12:15 – DEMO: Hyper-Personalized Lead Engagement Workflow\n • LinkedIn scrape → avatar extraction\n • Canvas-style image generator (Canva Enterprise features for free)\n • Auto-draft personalised Gmail invite\n24:30 – Deep-Dive: Canvas Image Personalisation – drag-and-drop fields & background tricks\n29:40 – Telegram Bot Integration (Start-to-Finish)\n • Creating the bot with BotFather\n • Setting webhooks & tokens\n • Wiring AgenticFlow triggers\n38:00 – Webhook Setup Walk-through – live terminal commands & common pitfalls\n45:45 – Two-Way Messaging – send & receive Telegram chats via workflow skill\n52:30 – Live Q&A\n • Model performance vs tools (why 9 tools/agent is sweet-spot)\n • Long-term memory & mem-DB hacks\n59:10 – Multi-Agent Beta Timeline – dev status, July roll-out plan, tier access\n1:05:25 – Database-vs-Knowledge Preview – upcoming Postgres-backed “fast tables” for 1M+ rows\n1:12:35 – Closing & Next Steps – where to grab the JSON files & join the community\n\n🔗 Useful Links\n• Discord community – https://qra.ai/discord\n• Product roadmap – https://agenticflow.featurebase.app/\n• Start building – https://agenticflow.ai\n\n👍 Like • Comment • Subscribe to stay in the loop!"}