ViaBTC CEO Haipo Yang: From Nof1 to x402 — A Look at AI Agent Applications and What’s Next

With Nof1’s stay AI buying and selling competitors and Coinbase’s newly launched x402 protocol changing into main trade speaking factors, AI Agents are quickly increasing their use circumstances throughout finance and funds.

As a consultant protocol for AI funds, how does x402 differ from conventional cost methods? What eventualities does it serve? And as AI funds mature, what different foreseeable functions would possibly AI Agents unlock? This Guest Expert piece summarizes views shared by Haipo Yang, Founder and CEO of ViaBTC, on the feasibility of x402 and the long run potential of AI collaboration networks.

Q: x402 has just lately grow to be a scorching matter within the trade. What is the view on utilizing token funds—like x402—to clear up cost issues for AI?

Haipo Yang: From an engineering standpoint, x402 is a comparatively easy protocol. Its core worth just isn’t inventing a brand new cost methodology, however packaging on-chain funds as a standardized net service—and introducing a Facilitator to tackle belief and execution challenges in on-chain funds.

Many comparisons are made between x402 and conventional cost methods, however these methods serve totally different “customers.” Alipay and Visa supply glorious cost experiences, however they’re designed for people, not for AI Agents. For AI Agents, conventional cost methods at present create two apparent obstacles:

1. 1) High entry limitations: It is tough for scripts to open financial institution accounts and full KYC, whereas producing a pockets tackle able to paying on-chain might be completed with a single line of code.
2. 2) High friction prices: AI interactions are high-frequency and fragmented. An Agent would possibly name a knowledge API as soon as and pay $0.0001. Routing that by way of card networks can introduce charges that exceed the cost itself.

In apply, x402 leverages token programmability—along with the middleman function of the Facilitator—to allow automated micropayments. In this context, the Facilitator features like “Alipay for the machine world,” absorbing on-chain affirmation complexity so Agents can full high-frequency transactions in milliseconds.

In standard on-chain funds, interactions might be gradual and complicated. x402’s method permits a Facilitator to function as an execution layer for on-chain transactions: verifying signatures, fronting gasoline, submitting transactions, and dealing with on-chain particulars. The payer submits a signature to the Facilitator slightly than straight performing on-chain operations. For each consumers and sellers, this reduces integration complexity by centralizing belief and settlement within the Facilitator.

Q: What is the outlook for x402, and what limitations would possibly it face in real-world adoption?

Haipo Yang: x402’s long-term worth primarily lies in an Agent-to-Agent financial community slightly than consumer-facing cost experiences. For finish customers, funds ought to grow to be invisible. In the long run, an AI Agent is unlikely to ask a consumer to “scan to pay.” Instead, a consumer would possibly set an instruction equivalent to “Analyze the market each morning at 9 a.m.” The Agent may then name a number of service suppliers within the background for information or social knowledge. Fees generated by high-frequency API calls might be settled robotically by way of x402, enabling service consumption end-to-end with minimal human intervention.

This mannequin can shift API monetization from subscription memberships to really pay-as-you-go utilization, as a result of x402 naturally suits machine-to-machine collaboration that’s high-frequency and extremely fragmented.

There can be an often-overlooked safety benefit. Allowing an Agent to transact utilizing a bank card quantity creates successfully limitless legal responsibility. If an Agent is compromised or behaves incorrectly, it may generate uncontrolled spending. With a token pockets, spending limits might be enforced—for instance, a capped “pocket cash” stability of 100 USDC—maintaining potential losses controllable.

However, x402’s simplicity additionally makes its limitations clear. The protocol depends closely on Facilitators equivalent to Coinbase. This simplifies growth however introduces a centralization threat and a possible single level of failure. If a Facilitator goes offline, behaves maliciously, or censors transactions, the cost stream can break.

In addition, as a result of x402 is designed to be easy, it doesn’t cowl sure real-world commerce necessities—equivalent to refunds—throughout the protocol itself. Disputes round unfinished companies or faulty items usually require reversals, and irreversibility could make such flows more durable to implement.

In parallel, broader Agent cost protocols are being explored, together with Google’s AP2, with objectives equivalent to accommodating card networks, supporting cryptocurrencies, and dealing with complicated flows like refunds. In the long term, extra complete requirements could also be fascinating—however multi-stakeholder complexity can gradual deployment. x402’s benefit is fast usability: a pockets plus code is adequate to begin.

Q: In apply as we speak, the place are AI Agents delivering actual worth?

Haipo Yang: At current, the largest beneficiaries of AI Agents stay builders. AI pair programming has grow to be routine for a lot of engineers, and instruments equivalent to Cursor have seen broad adoption. For giant, architecturally complicated tasks, full accountability is often not delegated to Agents at this stage. But for tedious, time-consuming duties—equivalent to code assessment, unit testing, and components of algorithmic logic technology—Agents can meaningfully cut back workload and save time.

Another notable space is enabling non-technical customers. “Vibe coding” has attracted consideration as a result of it permits individuals with out programming backgrounds to translate concepts into code by way of pure language. That stated, Agent output usually requires repeated debugging. Rapid prototyping turns into doable, however after many iterations codebases can grow to be bloated and more durable to preserve. Even so, a partial success price can nonetheless be beneficial as a result of it allows a 0-to-1 leap for non-technical creators.

Agents are additionally more and more helpful for small, widespread office wants. For instance, producing an icon, a button fashion, or a easy UI sketch beforehand required designer assist. Agents can now produce fast drafts, lowering back-and-forth and accelerating iteration.

Despite present limitations, these capabilities are already adequate for small groups and impartial builders constructing demos or MVPs.

Q: Looking forward, the place is the largest alternative for AI Agents—and may crypto see comparable new experiments?

Haipo Yang: Over an extended cycle, the chance for AI Agents is unlikely to stay confined to developer help. Future potentialities embrace extra autonomous collaboration and autonomous procurement.

Industry experiments are rising. For instance, Nof1’s stay AI buying and selling competitors successfully permits Agents constructed on totally different fashions to check technique capabilities in actual market environments. In this setting, Agents transfer past offering info to people and start forming closed loops of notion and motion.

More exchanges are additionally beginning to assist MCP (Model Context Protocol). CoinEx, throughout the ViaBTC ecosystem, has revealed an MCP service on GitHub. With MCP companies, an Agent can straight entry an alternate’s real-time quotes, candlestick (Ok-line) knowledge, and information feeds, then mix that knowledge with mannequin reasoning for deeper evaluation. In precept, an Agent can generate methods primarily based on a consumer’s threat preferences and—when deployed domestically—may place orders robotically.

This trajectory allows automated buying and selling and extra clever market making. By observing real-time market depth, volatility, and buying and selling quantity, an Agent can dynamically modify order costs and sizes, bettering market effectivity and liquidity. These developments point out a shift from “serving to with analysis” to “supporting choices and execution.”

Within this mannequin, x402 can present the financial rail for Agent collaboration. For instance, an Agent tasked with producing an in-depth Bitcoin analysis report might lack sure knowledge inputs. It can robotically name different Agents for on-chain place and transaction datasets, or for sentiment summaries aggregated from information, finishing micropayments for every service behind the scenes. The finish consumer receives a single report, whereas a number of Agent-to-Agent microtransactions happen within the background.

Taken collectively, Nof1 highlights decision-making in stay environments, MCP helps knowledge entry and execution, and x402 allows financial collaboration amongst Agents. As Agents grow to be able to find sources, buying companies, invoking instruments, and finishing full activity chains, the consequence more and more resembles a digital financial system composed of many cooperating Agents.

The put up ViaBTC CEO Haipo Yang: From Nof1 to x402 — A Look at AI Agent Applications and What’s Next appeared first on BeInCrypto.