The 3 key challenges for multi-agent system design according to Chi Wang of Google DeepMind: - Consensus - Shared state - Interop Protocol designers and devs anything sounds familiar?! The frontier of AI requires solving problems we’ve been facing for years in Ethereum. Lots

1/ Jolt just got a 6× speedup. We now prove over 1,000,000 RISC-V cycles/sec on a 32-core CPU and over 500,000 cycles/sec on a MacBook, with proofs of about 50 KB. Here’s what’s new 🧵

0/ What if Web3​ could superpower AI by enabling agent collaboration across organizations in untrusted settings? Our ERC-8004: 🤖Trustless Agents extends Agent2Agent protocol, providing a unified interface to discover agents and establish trust through reputation and validation.

MassGen 0.0.10 is out with Azure OpenAI backend support!  🚀 This is an important step in expanding MassGen’s support for enterprise users. Take it for a spin with Azure OpenAI and tell us how it works for you! github.com/Leezekun/MassG… #AI #AgenticAI #MultiAgent

1/ We made verifiable AI (zkML) 3-7x faster than everyone else! Here's how. We asked a simple question. What if we modified a16z Crypto's sumcheck based & lookup-centric zkVM "JOLT" for ML operations?👇🧵

Excited to see the ERC-8004 proposal by Davide Crapis and Marco De Rossi! Trustless agents are powerful thanks to their credible neutrality, transparency, and immutability. Sparsity and MassGen have been cooking, and we look forward to future collaboration!

When I worked at Snap Inc., I saw Evan Spiegel file 10+ bugs on Snapchat every single day. He used the product for hours every day. True product-oriented mindset in action. Still one of my most memorable lessons — and a north star as we build our own journeys.

For institutional DeFi adoption, what's the most critical bottleneck to solve FIRST?

Glad to see that NEAR Protocol is validating the private AI direction we pointed out 4 months ago. The problem is not privacy. It is user data ownership (UDO) which implies influence on who can view and monetize. UDO will be the biggest landslide change in Internet history🚀

During the collaboration with the Ethereum foundation on ERC-8004, I wrote a short recap of the TEE technology. Given the amount of attention on TEE right now in web3, I want to share it here in case it is useful for anyone else. docs.google.com/document/d/1k8…

We are glad to demo at the first ERC-8004 Trustless Agent call! Thanks to Davide Crapis, Marco De Rossi, and the whole 8004 team for organizing — love the energy in the group. 8004 let’s go!

Recently, Coinbase and Ethereum have unveiled infrastructure that fits together for a bigger vision: ETH as the default rails for AI agents to transact and collaborate seamlessly. Last week's x402 Foundation and Cloudflare's NET Dollar announcements seal the full stack,

Another step forward for DeFi. We’re teaming up with Sui to bring the first SEC-registered stablecoin, YLDS, natively to Sui. As part of this, DeepBook Protocol on Sui's new margin system will integrate YLDS to transform idle USDC into yield-bearing assets.

Tomorrow → live workshop on ERC-8004 (Trustless Agents) + verifiable compute ⚙️ hands-on with a free TEE lab env — just bring your laptop 💻 hosted by Sparsity (a16z-backed) luma.com/ezd2z3s8

Founders often ask me what I've done to build a social media following. It's actually quite simple, and boils down to one piece of advice: Be vulnerable. That thing you think you're not supposed to tweet? Tweet it. The story that makes you look weak or insecure? Share it

We’re excited to announce MassGen v0.0.32 🚀, now with support for command-line execution in Docker! Each agent can run commands in its own isolated Docker environment, eliminating dependency conflicts and ensuring smoother collaboration across projects. Give it a try and

People may not be aware that you can pay gas fees on Base with any ERC20 token (including USDC). Link below for more info. Try it out!

TUMIX: Multi-Agent Test-Time Scaling with Tool-Use Mixture Tomorrow PST 9am meet.google.com/yzs-rpgi-chc?a… Summary: HLE 21.6->34.1 on Gemini-2.5-pro via diverse tools (text-only, code, search, etc.) mixing strategies Speaker: Yongchao Chen, final year MIT PhD

🚀

Honored to have such an awesome community! Keep building💪