کمپانی databricks یکی دو روز پیش Omnigent رو اپن سورس کرده و به نظرم یکی از جالبترین لایبرری ها در دنیای Agentهاست.
بیشتر آدمها و تیمها هنوز روی ساختن Agentهای بهتر تمرکز دارن. اما به نظر من فرصت بزرگتر، ساختن Harnessهای بهتره.
بهترین تیمها الان از چندین Agent مختلف استفاده میکنن؛ Claude Code، Codex، Gemini، ابزارهای داخلی و Agentهای اختصاصی. مشکل اصلی دیگه توانایی مدلها نیست، بلکه هماهنگ کردن و مدیریت این Agentهاست.
چیزی که در Omnigent برام جالبه اینه که خودش یک Meta-Harness هست؛ یعنی لایهای بالاتر از Agentها که وظیفه مدیریت، هماهنگی، حافظه، قوانین و همکاری بین اونها رو بر عهده میگیره.
به نظرم مسیر تکامل این فضا چیزی شبیه اینه:
Model → Agent → Harness → Meta-Harness
به نظرم آینده متعلق به یک سوپر Agent نیست. بلکه مجموعهای از Agentهای تخصصی هست که داخل یک کنترل پلین مشترک کار میکنن.
همونطور که Kubernetes جای Containerها رو نگرفت و فقط لایه مدیریت اونها شد، Meta-Harnessها هم ممکنه به لایه مدیریت Agentها تبدیل بشن.
در نهایت برندهها لزوماً اونهایی نیستن که بهترین LLM رو دارن. بلکه اونهایی هستن که بهتر از بقیه میتونن مدلها، ابزارها، حافظه، Workflowها و انسانها رو در یک سیستم منسجم کنار هم قرار بدن.
این بلاگشون را بخونید که خیلی قشنگ توضیح میده. اگه هم هنوز keynote شون را ندید حتما ببینید.
Blog: https://t.co/7Qj98FvQcT
Youtube: https://t.co/18Gr9C0f3p
A $500B blockchain exists because a 15-year-old got upset over a World of Warcraft update.
Back in 2010, Blizzard nerfed Vitalik Buterin's favorite Warlock spell.
In his own words:
"I cried myself to sleep, and on that day I realized what horrors centralized services can bring."
He quit the game.
A year later, his father introduced him to Bitcoin.
He couldn't afford to buy any and couldn't mine it either.
So he started writing for a crypto blog that paid 5 BTC per article, worth roughly $3.50 at the time.
Today, those same 5 BTC would be worth over $500,000.
At 17, he co-founded Bitcoin Magazine.
He later applied for a job at Ripple.
They wanted him, but couldn't sponsor his visa.
At 19, he proposed major upgrades to Bitcoin.
The community rejected them.
So he decided to build something new.
He wrote a whitepaper and called it Ethereum.
Within weeks, around 30 developers had reached out to help.
He dropped out of college, received a $100,000 Thiel Fellowship, and launched Ethereum's ICO at $0.31 per ETH, raising $18 million.
One participant bought 1 million ETH for $310,000.
At peak, that position was worth roughly $4.3 billion.
Another spent just $263 for 850 ETH.
Those coins are now worth well over $1.7 million.
Ethereum launched when Vitalik was only 21.
Peak market cap: over $500 billion.
All because Blizzard nerfed a Warlock spell.
🚀 اگر دوست داری Claude برات مثل یه Senior Software Engineer واقعی کد بزنه (نه فقط vibe coding)، این مجموعه Skills از Matt Pocock رو بهش بده!
مجموعهای از مهارتهای حرفهای:
Architecture
TDD
Refactoring
Code Review
و مهارتهای سطح بالا
https://t.co/OTxL7e7KQR
Who is behind @solana?
you probably already know toly, raj, lily and vibhu, but many other people shape solana every day:
• @lagunacarta: Ecosystem Marketing
• @niran7: Head of Owned Media / Marketing
• @nickducoff: Head of Institutional Growth
• @stevensarmi: Engineering
• @therealchaseeb: Founders & Beyond
• @jnwng: Vice President of Product
• @CaddleMaya: BD Payments
• @Pedromiranda: Consumer Growth
• @ramzyyalii: DeFi Growth
• @AntonioNetoSOL: LATAM Growth
• @catgu_: Head of Digital Assets Product
• @stokenomic: Recruiter / HR
alongside many other legends building solana as we know it today ⚡️
①
شرکت #Anthropic ۵ دوره رایگان درباره Claude و MCP داده:
📚 ۲۸۷ ویدیو
🎥 ۲۶ ساعت آموزش
🧪 ۳۳ تمرین
موضوعات: ClaudeAPI، #MCP، #Amazon#Bedrock، #Vertex #AI# و MCP پیشرفته.
Introducing the Phantom MCP Server 🦞
Agents can swap, sign, and manage addresses across all of Phantom's supported chains.
Ready to work with Claude, OpenClaw, or any MCP-compatible client.
Stop prompting AI to "make me a landing page."
200+ hours testing taught me this:
Generic prompts → pretty pages
My prompts → pages that sell
The pack includes:
→ 3D Landing Hero Section
→ Sentient 3D Core
→ Cyberpunk Volumetric Shaders
→ Microlender Protocol
→ Hyper-Futuristic MCP-2099
Like + reply "PROMPT" for the library.
(following required for DM)
Chinese engineers literally refactored @OpenClaw in Go for insane efficiency 🤯
It runs on a $10 @Raspberry_Pi instead of a $399 Mac mini.
→ Uses <10 MB of memory (99% smaller than OpenClaw)
→ 400× faster startup time
→ Boots in 1 second
Free and open source, repo in 🧵↓
OpenClaw/Clawdbot masterclass!
In this video, I give you the complete blueprint to get up and running with your own Clawdbot. And not just one, but multiple AI employees working 24x7 for you.
Enjoy!
Chapters:
00:00 - Intro
1:12 - See my 3 AI employees in action
5:49 - How the whole thing works under the hood
9:47 - Installation and setup
11:22 - Hooking Clawdbot to Telegram
15:35 - Understanding the project structure
20:03 - Adding custom skill (from a repo of 42k skills)
23:18 - Automating your bots with Cron jobs
25:07 - Going from 1 to 10 Clawdbots
28:53 - Cron jobs vs. Heartbeats: what's the difference?
30:36 - What's coming up next?
31:34 - Outro
➥ The Agent GDP Economy Infrastructure
Passive chatbots are dead. What's replacing them? Autonomous AI agents that own wallets, launch tokens, and execute trades without asking permission.
The OpenClaw ecosystem runs on @Base: Coinbase's L2 where gas costs pennies and AI agents operate as first-class economic actors.
The Core Stack
Four projects form the financial backbone of autonomous agents:
❶ @clanker_world / $CLANKER → Token factory. Agents spin up their own tokens in seconds, building economies around memes, predictions, or pure speculation. e.g.: @Clawnch_Bot
❷ @bankrbot / $BNKR → The agent's bank account. Cross-chain DeFi, automated trading, yield farming. If an agent needs financial rails, Bankr provides them.
❸ @flaunchgg / $FLNCHY → CLI launch platform. Command-line token deployment for AI projects. Powers platforms like @Moltlaunch where agents bootstrap economies at terminal speed.
❹ @virtuals_io / $VIRTUAL → Full-stack monetization layer. Tokenizes agents themselves via Agent Commerce Protocol. Shared ownership, on-chain revenue splits, automated buybacks. Already processing 100k+ transactions.
— Why This Matters
These aren't separate tools. they're Lego blocks for agent economies:
▸ Clanker + Flaunch = instant token launches
▸ Bankr = the treasury managing those tokens
▸ Virtuals = turning the agent itself into an investment vehicle
▸ Other supporting infra = @xmtp_ (messaging) + @neynarxyz (social graphs) = communication layer
The result?
Agents that launch tokens, manage communities, execute trades, and distribute profits. all programmatically, all on-chain.
Base ecosystem is thriving with frameworks like @Moltbook and @openclaw are already seeing activity spikes. competition is heating up (Bankr vs. Clanker in launch volume) and Virtuals is running 60-day builder programs.
Bottom line, this isn't a sandbox experiment.
Base's low fees make micro-transactions viable. open-source frameworks mean anyone can fork and deploy. and speculation on AI-as-economic-participants is real capital flowing into real infrastructure.
—
cc @jessepollak