There is a strange phenomenon in the crypto world: everyone is trading Bitcoin and Ethereum, but the stablecoins that are actually used the most in daily life are gaining traction, especially for cross-border payments. Stablecoins have become the first choice for many. However, if you observe closely, you will find that the use cases for stablecoins are still primarily human-driven, with almost no cases of AI agents using stablecoins. What KITE wants to change is exactly this.
First, let me mention a detail that many people have overlooked: the stablecoins chosen by KITE are USDC and pyUSD, not USDT. This choice is quite significant. Although USDT has the largest market capitalization and the best liquidity, its transparency has always been questioned. Are the reserves sufficient? How reliable are the audit reports? These issues have not been thoroughly resolved until now.
USDC is issued by Circle, a regulated U.S. company, with reserves consisting of U.S. Treasury bonds and cash, and monthly audit reports. This level of transparency is unmatched by USDT. More importantly, Circle is backed by traditional financial giants like Goldman Sachs and BlackRock. If regulations tighten in the future, USDC's chances of survival will far exceed those of USDT.
pyUSD is even more interesting; this is the stablecoin issued by PayPal. PayPal operates at a massive scale, with over 400 million users and processing tens of millions of transactions daily. If pyUSD can be integrated with PayPal's payment network, then this use case is incomparable to any other stablecoin.
In KITE's investors, there happens to be PayPal Ventures. This is not a coincidence but a strategic layout. PayPal's investment in KITE clearly indicates that they see the long-term value of AI Agent payments, and PayPal itself has pyUSD, which completely aligns with KITE's technological direction. In the future, there will likely be deeper integrations.
Imagine a scenario where you shop on Amazon or Shopify, but it's not you placing the order; it's your AI assistant Agent helping you. It knows your coffee at home is running low and automatically compares prices across several e-commerce platforms to find the best deal, then pays directly with pyUSD. You don't need to participate in the entire process; you only receive a notification on your phone that the order is complete.
The realization of this scenario requires several preconditions: Agents must have payment capabilities, possess a trustworthy identity, be able to prove they have permission to spend your money, and the payment process must be secure and auditable. In case an Agent purchases the wrong item, you should be able to trace the entire process. These conditions are unattainable in traditional payment systems, but KITE can achieve them.
KITE's three-layer identity system is specifically designed for this scenario. The user layer uses your EOA account, which holds the highest authority, akin to your main wallet. The Agent layer uses sub-identities derived from BIP-32, which have limited permissions, such as a maximum expenditure of $100 per day and restricted to specific merchants. The session layer uses temporary keys, generating a new key for each payment and destroying it after use.
The most clever aspect of this design is that even if the Agent's session key is compromised by hackers, the hacker can only execute one transaction and must do so within the permission limits. Exceeding these limits would result in an automatic rejection by the smart contract, making losses controllable. This is a practical application of the Bounded Loss Theorem.
In the bi-weekly AI industry report released by KITE on December 15, it was mentioned that both Visa and AWS are working on agentic commerce infrastructure. This signal is significant, indicating that traditional payment giants have also recognized the potential of the Agent economy and are beginning to lay out their plans.
Visa has been doing payments for decades, processing tens of thousands of transactions per second globally, with a deep technological accumulation. If they fully enter the Agent payment space, what advantages does KITE have? The core advantages are decentralization and verifiability. Visa's system is centralized; you must trust Visa as the intermediary. But KITE is decentralized, requiring no trust in any party. All rules are written in smart contracts and executed automatically on-chain.
Moreover, Visa is constrained by regulation and compliance, making it hesitant to touch many innovative payment scenarios, such as cross-border micropayments or automatic settlements between Agents. These are difficult to implement within the traditional financial system, but KITE, as a blockchain project, can explore more possibilities within the compliance framework.
From a business logic perspective, the payment needs of AI Agents are completely different from those of humans. A person might swipe their card dozens of times a day, but an Agent might need to process hundreds of micropayments every second. Humans can wait a few seconds for confirmation, but Agents require millisecond-level responsiveness. People are relatively insensitive to transaction fees; swiping can incur fees of 1-3%, but for Agents handling micropayments, if the fee exceeds 0.1%, it generally becomes unprofitable.
KITE's state channel technology specifically addresses this problem. The opening and closing of channels happen on-chain, while thousands of transactions in between occur off-chain, achieving speeds in the millisecond range with latencies below 100 milliseconds and transaction fees as low as $0.000001. This cost is practically negligible.
Moreover, state channels have three types: unidirectional for continuous payment scenarios, such as subscription services; bidirectional for mutual settlements between Agents; and programmable for complex conditional payments, like 'payment only occurs when a specific event happens.' This flexibility is unattainable in traditional payment systems.
McKinsey estimates that by 2030, the economic scale of AI Agents will reach $4.4 trillion. If 10% of these transactions are settled with stablecoins, that would amount to $440 billion in annual transaction volume. If KITE captures 1% of this market, that would translate to $4.4 billion in annual transaction volume. With a fee rate of 0.1%, that equates to $4.4 million in annual revenue, and this is just a conservative estimate.
But to be honest, whether this market can really take off depends on several key issues. First is regulation; the regulation of stablecoins is still unclear globally. The U.S. is pushing MiCA while the EU is also formulating its own framework. If regulation tightens, the use cases for stablecoins will be limited. Second is user acceptance; although AI Agents are cool, most people are still accustomed to handling things themselves. How many are willing to hand over their wallets to Agents? Third is technological maturity; KITE is still in the testnet phase, and the mainnet won't go live until Q1 2026. Whether any issues arise during this period is uncertain.
However, from KITE's preparatory work, they have considered these issues. Regarding regulation, they proactively designed a compliance module that can incorporate optional KYC/AML at the application layer, preserving the underlying permissionless nature while meeting regulatory requirements in various regions. Regarding user acceptance, they chose to collaborate with giants like Shopify and PayPal to directly gain access to existing users and use cases. In terms of technological maturity, the testnet has been running for nearly a year, with total transactions exceeding 500 million, and basic functionalities have been validated.
What's more critical is that KITE is not betting on AI Agents completely replacing humans, but rather on Agents becoming assistants to humans, handling those repetitive, low-value, yet necessary tasks. For instance, renewing subscriptions monthly, restocking periodically, or automatically filing taxes. Using Agents for these scenarios will significantly increase efficiency without requiring users to change many habits.
From the development trend of stablecoins, the global market value of stablecoins has now exceeded $180 billion and is growing rapidly, especially in emerging markets. In many countries, local currencies are unstable and inflation is severe, leading locals to prefer holding dollar-pegged stablecoins. This demand is real.
The emergence of AI Agents will greatly expand the use cases for stablecoins because Agents don't care about exchange rate fluctuations or the cumbersome processes of cross-border transfers. They only care about efficiency and cost. The 24/7 trading of stablecoins without intermediaries and confirmations in seconds are essential features for Agents.
The logic behind PayPal issuing pyUSD is similar. They know it's challenging to promote stablecoins on a large scale purely through B2C. However, if they focus on B2B, allowing enterprises to use Agents for payments, the use cases open up completely. An e-commerce platform might have millions of merchants. If these merchants use Agents to handle orders, manage inventory, and automatically restock, the transaction volume would be astronomical.
What KITE aims to do is provide the infrastructure for the B2B market, not to educate consumers about stablecoins, but to enable enterprise Agents to seamlessly use stablecoins. Merchants do not need to understand what blockchain or private keys are; they only need to know that Agents can automatically complete payments and that the cost is lower than credit cards and faster than bank transfers. That's enough.
From the testnet data, this direction has been validated. There are over 74 million unique addresses and 1.7 billion Agent calls. These numbers prove that developers recognize KITE's technical solution and are actively trying various application scenarios. Once the mainnet is launched, these test applications can be directly converted into commercial applications, immediately generating transaction volume.
Lastly, one point that many people overlook is that KITE's choice of stablecoins has a strategic consideration, which is to avoid direct competition with Bitcoin and Ethereum. If KITE develops into a universal public chain, it would have to compete with Ethereum for DeFi, with Solana for high performance, and with various L2s for ecosystems. This red ocean is too crowded.
However, if they focus on Agent payments, using stablecoins as the native token, that represents a relatively blue ocean market with few competitors and clear differentiation. This positioning is very clear; it’s not about doing everything, but rather about capturing a niche segment and executing it to perfection.
When both PayPal and Visa are focusing on the AI Agent wallet, KITE has already set up the infrastructure. Next, it depends on who can capture the market faster, who can provide a better developer experience, and who can integrate more application scenarios. This war has just begun.
