Good Morning ☕️

The markets continue their downward slide with BTC briefly heading below $80K, and the AI agent sector seems to be in free fall with Virtuals Protocol's revenue crashing 97%. But as Warren Buffett famously said, "Be fearful when others are greedy, and greedy when others are fearful." While sentiment tanks, we're seeing the strongest projects continue to build.

This Week:

• OpenAI releases GPT-4.5 with enhanced EQ and creativity

• Virtuals Protocol revenue collapses

• CZ joins Vana as advisor following YZi Labs investment

• The "uncensored" vs "permissionless" debate

• Podcast with Sapien

Let's dive in... 👇

Highlights

🔥 Virtuals Protocol Revenue Crashes over 97% since January’s peak, with daily fees plummeting from over $1 million to under $30,000 across all networks.

The collapse in trading activity and new agent creation (now below 10 daily) shows the extent of the correction in the AI agent space, challenging the protocol's viability during periods of reduced hype.

🔥 OpenAI has released GPT-4.5 as a research preview, offering their "largest and best model for chat yet" with improved factual accuracy and significantly reduced hallucinations.

The model showcases enhanced emotional intelligence ("EQ") and better understanding of human intent, making conversations feel more natural and intuitive while demonstrating stronger aesthetic intuition and creativity, capabilities that language models have historically struggled to master.

🔥 CZ Joins Vana as Advisor Following YZi Labs (formerly Binance Labs) Investment, bringing his experience and aligning his reputation with Vana's blockchain for AI data sovereignty as part of YZi Labs' strategic push beyond Web3.

The partnership aims to accelerate Vana's ecosystem of 16+ DataDAOs, positioning users to capture value from their personal data in the estimated $1 trillion data economy while providing AI developers access to higher-quality training datasets.

Uncensored Tech Debate

Since it was released a couple of weeks ago, Grok 3, branded as an "anti-woke" uncensored AI assistant, has sparked controversy by generating "execution target" suggestions and instructions for illegal activities. While xAI blamed a "rogue engineer," the incident raises a familiar question for those of us involved in crypto:

What does it mean for a technology to be "uncensored"?

After the Bybit hack, we have seen some similar debates happening in crypto with calls for “chain rollbacks” and DeFi protocols facing choices about allowing stolen funds through their systems. Some maintained complete neutrality, while others disabled front-ends or suggested protocol interventions.

Venice founder Erik Voorhees articulated this tension in a recent tweet:

There’s an interesting parallel between the situation with Grok and illicit funds in crypto: crypto is designed to be permissionless by architecture; neutral protocols that process all valid transactions without judgment. AI systems like Grok, however, are inherently centralized; their "uncensored" nature is simply a policy choice that can be modified at any time.

Yet both technologies grapple with the same fundamental question: should these systems operate neutrally regardless of how they're used, or should certain behaviors be restricted?

In crypto, true permissionlessness prevents censorship by design, while AI's "freedom" exists only at the discretion of its creators.

As Voorhees notes, "Being decentralized exists on a gradient, being permissionless is not." Perhaps AI will never be truly permissionless, but both technologies force us to decide: do we value absolute neutrality over potential misuse?

What do you think?

Podcasts & Streams

New Outpost Episode: How Sapien Lets Anyone Earn by Creating Datasets

This week, we sat down with Trevor, co-founder of Sapien, to explore how the global data economy is evolving to fuel better AI models. Trevor breaks down why Fortune 500 companies are scrambling to build vertical AI models with proprietary data, how decentralized networks can connect specialized experts to high-value labeling opportunities, and why the "data wall" for openly available internet content is forcing AI companies to find new data sources.

For anyone curious about how humans will earn a living in the AI economy, this conversation offers a fascinating glimpse into what the future might look like.

Other Great Listens:

• Chris Dixon and David George from a16z discuss how "AI has broken the economic covenant of the internet" and the intersection between AI and crypto.

• Teng Yan joins Grant on the Blocmates podcast to detail the current state of crypto and AI markets, covering the trajectory of AI agents, the importance of crypto verification for AI systems, and why infrastructure projects like Mira Network, Nillion and Automata will be critical in 2025.

Project Updates

📍 Orbit: Released V2 trading AI, improving risk management strategies in DeFi applications.

📍 MyShell: Announced ShellAgent Protocol updates, integrating IM and on-chain intelligence, improving AI-assisted trading.

📍 Akash Network: Deployed ElizaOS AI agents on the Akash Supercloud, enabling instant AI deployment and integration through their API for decentralized compute resources.

📍 NetMind AI: Launched a no-code AI agent creation tool on BNBChain to simplify AI adoption for Web3 builders while unveiling Zoey Well-Being Agent, an AI designed for personalized longevity insights.

📍 NEAR Protocol: Partnered with Eliza OS to offer developers free AI inference via NEAR AI Hub, plus agents capable of handling transfers, swaps, and tracking across NEAR, Bitcoin, Cosmos, and EVM chains.

📍 Allora Network: Dev mainnet is now live. Also collaborated with PolyhedraZK to integrate zkML with federated learning.

📍 Flock IO: Released HeyAni AI Crypto Agent, utilizing federated learning for real-time tokenomics analysis.

📍 GenSyn: Released Verde, a high-scale verification system for machine learning over untrusted nodes, enabling reliable onboarding of distributed compute devices worldwide to lower AI training costs.

📍 SentientAGI: Unveiled Dobby 70B, a community-owned AI model that advances decentralized AI governance and opened a waitlist for Sentient chat.

📍 Ungate: Deployed a decentralized execution cloud for AI agents, deployed as an Eigenlayer AVS, ensuring secure and transparent agent operations.

📍 Spheron: Locked in a strategic funding round of an undisclosed amount.

📍 Navigate AI: Launched an AI data marketplace on Base.

📍 PondGNN: Expanded their onchain machine learning for wallet profiling to help track wallets linked to the Bybit hack.

📍 ionet: Reported a new all time high revenue milestone, achieving over $65k in total network earnings in a 24hr period.

📍EMCprotocol: Deployed their mainnet for decentralized AI compute infrastructure.

Reading List

📚️ Deep reading, deep research, & deeply personal by YB—A candid exploration of how AI is changing the way we learn and create content, with practical insights on using tools like OpenAI's deep research to accelerate learning. YB shares his journey toward becoming a crypto x AI expert and why evocative, personal writing will remain valuable even as AI content generation improves.

📚️ AI Agents Bottom? What You Need to Know by Ailea Bulletin—An analysis of the recent market correction in AI agent tokens and why infrastructure protocols like Nillion and Bittensor are gaining attention as the focus shifts from model development to applications. The piece suggests we may have seen a bottom in the agents sector at $6B market cap, with several tokens showing resilience despite broader market weakness.

📚️ Dynamic TAO: Your No-Nonsense Guide by Teng Yan—A comprehensive breakdown of Bittensor's new dynamic TAO system and how it's transforming the network's token economics. The article explains the shift from validator-controlled emissions to a market-driven approach using subnet Alpha tokens, providing investors with a clear timeline of what to expect as the system evolves over the coming months.

Tools

Finding alpha in the current fast-paced AI meta requires the right tools. Here are some of the platforms we're using daily:

🛠️ AI Agent Toolkit: Comprehensive directory of AI agent resources, from frameworks and launchpads to tutorials and infrastructure

🛠️ Cookie.fun: Track agent metrics, market caps, and mindshare in real-time

🛠️ Kaito.ai: Using for mindshare and smart followers tracking on X. Kaito is conducting an airdrop using “Yaps” points from activity on X, connect your X account to participate here.

🛠️ Taostats.io: New analytics dashboard for monitoring Bittensor network activity and performance

🛠️ The Veldt: An AI copilot tailored to help you find alpha in the crypto trenches. We’re a big fan of the AI feed!

Feedback

We listen and improve. Have more feedback? Reply to this email!