Disclosure: The views and opinions expressed here belong solely to the author and do not represent the views and opinions of crypto.news’ editorial. AI agents are on the way. In these first few months of the year, there’s a wave of investment and excitement in this space, with autonomous assistants promising to take on more personal and professional tasks. From trading partners to science researchers, the applications are expansive. Read more: AI agents: The key to blockchain’s future or its biggest risk? | Opinion But there’s a critical gap between promise and performance—agents are only as capable as the data they can access. Often, this data is pulled from a patchwork of siloed, unverified, or public sources. Worse, there’s no hard and fast infrastructure to ensure private and secure processing. Therefore, agents need better data “diets” to overcome the expectation gap, achieve contextual understanding, and deliver instantaneous accuracy. The good news is blockchain offers a solution—backends where agents can access verified data and securely process it through trusted execution environments. Combined with on-chain insights, we can deliver a one-two punch of better decision-making and stronger privacy. The result? Agents that truly know you and, importantly, are worthy of your trust. Garbage in, garbage out Agents are like any technology in that input determines output. But right now, we’re facing a “garbage in, garbage out” problem. We’re feeding these models with copious amounts of information, but—without set ways to verify, authenticate, or meaningfully connect it—they’re struggling to put it together. There’s little chance of agents hitting the mark if the data coming in isn’t qualitative and the framework powering them isn’t defined. In this sense, agents are like us. If we only eat junk food, we’re slow and sluggish. If agents only consume copyright and second-hand material, they’re inaccurate, unreliable, and general rather than specific. Their data “diet” determines performance. Take a trading bot, for instance. Real-time access to market data, historical trends, and cross-chain activity is needed to make effective decisions. But when this data is scattered across the ether or inadequately verified, the bot’s decision-making is limited. The solution isn’t just more data—it’s better data. These emerging platforms need infrastructure to transform raw information into actionable insights while maintaining strict privacy standards. We need a system where data isn’t just abundant but enriched with actionable metrics and validated through trustless consensus. Only then can agents move from imitating knowledge to independent reasoning. Improving data diets with blockchain Our industry needs to ask and answer some data questions before agents hit the mainstream: Where is the information coming from? Do we have creator consent and compensation? And how do we guarantee privacy during processing? In all three areas, blockchain can play a role in ushering in this new agentic era. For data provenance, blockchain offers newfound transparency. Every piece of information can be tracked to its source, creating an immutable record of origins and history. Through trusted execution environments and cryptographic proofs, agents can verify the authenticity of their sources and ensure they’re working with zero-party data rather than second-hand approximations. The question of creator consent and compensation also finds a natural solution in blockchain’s programmable nature. Users can bind their online identities to NFTs—like ERC-7231 —which act as digital passports. This standardization gives users better control over their data and the decision of whether or not to share it. If they choose to do so, users can open avenues for automatic compensation whenever this information helps train or improve AI. This signals an overdue shift from the old data extraction model to one of fair value exchange. Finally, blockchain architecture provides the security guarantees needed for safe processing. Agents can analyze information through chain ecosystems and zero-knowledge proofs without exposing sensitive details. This creates an ecosystem where high-quality insights flow freely, but private data remains protected. This is vital in building a key element for adoption: trust. A better diet, a better agent We need agents at their best now. They’re on the cusp of becoming our digital best friends and confidants. If they’re going to analyze our most sensitive information and suggest the next best steps, agents need a framework that bakes in trust. This is, again, the kind of traceability and trustworthiness that blockchain can offer. Emerging frameworks make this possible by layering verification mechanisms directly into agent foundations. As a result, agents are armed with both on-chain and off-chain insights—transaction histories, user profiles, and advanced metrics—while verifying authenticity through multi-step analysis. When an agent accesses your financial data or health records, this foundation guarantees your information remains protected while enabling the agent to develop human-like reasoning. AI agents are the informational version of “you are what you eat.” It’s up to us—the system makers and data connectors—to ensure they’re ready for mainstream adoption with strict data principles. Only then, by enjoying better data diets and digesting them effectively, can agents reach their potential of becoming our trusted superhuman helpers. Read more: Agents present a Netscape moment for AI and crypto | Opinion Author: Yukai Tu Yukai Tu is the chief technology officer at CARV. Yukai is an expert in confidential computing and blockchain and holds an MS in Computer Science from UCLA. At CARV, Yukai is helping build CARV SVM Chain and CARV’s D.A.T.A. Framework, an agentic infrastructure that extends SVM’s capabilities onto Ethereum and empowers AI agents with high-quality on-chain and off-chain data. He’s also worked as a software engineer at Google and Coinbase, a contributor to the Cosmos SDK, and a blockchain engineering lead at LINO Network.
