Without Agent Oracle, the AI economy is just a castle in the air.

In the past few months while working on the Agent system, I have become increasingly aware of something that has been seriously underestimated by everyone: no matter how powerful LLM becomes, it cannot truly and reliably assess the state of the real world. Once an Agent enters the actual execution layer—opening accounts, trading, accessing websites, submitting forms—it is exposed to a very high level of vulnerability. Because it lacks a “reality layer.” What we lack is an Agent Oracle, which is almost the cornerstone of the entire Agent ecosystem but has long been overlooked.

Why is LLM not enough? Because the essence of LLM's capability is to generate the most probable text, rather than being a system that infers the truth of the world. It does not verify the truthfulness of news, does not identify phishing links, does not determine whether a certain API is compromised, does not understand whether a regulation is truly effective, and cannot accurately grasp the real bias behind Powell's speech. All of these fall under “fact verification” rather than “language prediction.” Therefore, LLM itself can never become the “source of truth” for an Agent.

Traditional Oracles can no longer solve this problem. They excel at price truths: ETH/USD, BTC/BNB, indices, foreign exchange, on-chain TVL, and other structured, quantifiable, observable data. However, the reality faced by Agents is entirely different: unstructured events, multi-source conflicts, semantic judgments, real-time changes, and fuzzy boundaries—this is event truth, which is an order of magnitude more complex than price truth. Event truth ≠ price truth; the mechanisms of the two are completely different.

The event verification market proposed by Sora is currently the closest attempt in the right direction. The core shift of Sora is: truth is no longer generated by node voting, but rather by Agents executing real verification tasks. A query will go through data fetching (TLS, Hash, IPFS), outlier filtering (MAD), LLM semantic verification, multi-Agent reputation-weighted aggregation, reputation updates, and challenge penalties. Sora's key insight is Earn = Reputation: income comes from reputation, and reputation comes from long-term real work, rather than stake or self-assertion. This direction is very revolutionary, but it is still not open enough—the experts in real-world event verification are extremely diverse, ranging from finance, regulations, healthcare, multilingual, to security audits, fraud detection, on-chain monitoring, and industry experience, with no single team able to build an Agent cluster that covers all areas.

Therefore, what we need is an open, multi-agent participatory “truth game market.” Why? Because the way humans obtain truth is not by asking a single expert, but by checking multiple sources, asking multiple friends, listening to multiple KOLs, and then extracting a stable understanding from the conflicts. The Agent world must also evolve along this mechanism.

The direction we are building is a combination of ERC8004 + x402. ERC8004 is responsible for establishing a programmable reputation layer that records the historical performance, call frequency, success cases, challenge records, areas of expertise, stability, etc., of each Agent, allowing the “verifiable career” to naturally determine the eligibility of Agents to participate. On the other hand, x402 is responsible for the payment layer, through which we can dynamically gather multiple high-reputation Agents in a single event verification, allowing them to validate in parallel, cross-check, and aggregate output results based on contribution weighting. It's not about finding one expert, but rather convening a committee — this is the “truth committee” of the machine world.

An open, multi-agent, reputation-weighted, challenge-incentivized, and self-evolving truth market might be the true form of the future Oracle.

At the same time, Intuition is building another layer: Social Semantic Truth. Not all truths can be derived through event validation, such as “Is a certain project trustworthy?”, “Is the governance quality good?”, “Does the community like a certain product?”, “Is a certain developer reliable?”, “Is a certain viewpoint recognized by the mainstream?” These are not Yes/No questions, but rather social consensus, suitable for expression using the TRUST trinity (Atom — Predicate — Object), and consensus strength is accumulated through stake support or opposition. It applies to long-term facts such as reputation, preferences, risk levels, labels, etc. However, their current product experience is indeed poor; for instance, to create “Vitalik Buterin is the founder of Ethereum”, all related terms must have identities within the system, and the process is very awkward. The pain points are clear, but their solution is currently not good enough.

Thus, the future truth structure will present two complementary layers: event truth (Agent Oracle) responsible for real-time world, and semantic truth (TRUST) responsible for long-term consensus, together forming the truth base of AI.

The Reality Stack will be clearly divided into three layers: the Event Truth Layer (Sora / ERC8004 + x402), the Semantic Truth Layer (TRUST), and the final Settlement Layer (L1/L2 blockchain). This structure is likely to become the true foundation of AI × Web3.

Why will this change the entire internet? Because today's agents cannot verify authenticity, determine sources, avoid fraud, prevent data pollution, undertake high-risk actions, or cross-check like humans. Without an Agent Oracle, the agent economy cannot be established; but with it, we can finally create a verifiable layer of reality for AI. Agent Oracle = the reality foundation for AI.

The future Oracle will not be a node network, but rather composed of countless specialized Agents: they accumulate reputation through income, participate in verification through reputation, acquire new work and challenges through verification, collaborate automatically, divide labor automatically, and evolve themselves, ultimately expanding into all fields of knowledge. It will be a truly machine society truth market.

Blockchain provides us with a trustworthy ledger, while the era of Agents requires credible reality, credible events, credible semantics, credible judgments, and credible execution. Without Agent Oracle, AI cannot safely operate in the world; with it, we can establish the “reality layer” for machines for the first time. The future belongs to those protocols that can help machines understand the real world.