Author: Zen, PANews

With the rapid development of artificial intelligence, shopping and payment methods are being reshaped.

In April this year, Visa launched Visa Intelligent Commerce, using AI to connect the "from search to purchase" scenario, and cooperated with industry leaders such as Anthropic, Microsoft, Mistral AI, Stripe, etc., aiming to achieve personalized and secure AI commerce on a global scale.

Last month, Google announced a new AI agent for basic service tasks - its design covers restaurant reservations and will gradually expand to local service reservations and event ticketing.

Today, traditional giants are vying for the opportunity to establish AI agents as the next generation of mainstream user interfaces, extending their reach into the blockchain and cryptocurrency sectors. Earlier this month, Kite announced the completion of an $18 million funding round, bringing its total funding to $33 million. The project builds a trusted transaction layer for the agent economy, enabling agents to independently transact, coordinate, and operate. The platform aims to provide autonomous agents with encrypted identities, programmable permissions, and native access to stablecoin payments.

Unlike most Web3 projects, Kite counts several heavyweights from traditional industries among its investors—lead investors PayPal Ventures and General Catalyst, with participation from Samsung, 8VC, and SBI. So, why did so many leading institutions choose Kite?

Building native economic infrastructure for AI agents

Currently, most autonomous brokers are still deployed on centralized platforms, which are designed and optimized with human operators at the core. While this offers advantages in terms of ease of use, it forces brokers to rely on sometimes fragile authentication, authorization, and settlement processes, leading to efficiency bottlenecks and systemic risks.

In theory, existing blockchain infrastructure offers significant advantages over traditional payment methods, including immutable logs, cryptographic proofs, and replicable smart contract logic. Furthermore, blockchain-based payments can eliminate intermediaries and enable cross-border micropayments.

However, traditional blockchains, like Web2, are similarly user-centric and lack native identity and trust mechanisms for autonomous agents. Within traditional infrastructure, AI agents often "borrow" human identities to operate, leading to identity fragmentation and security risks (an M×N verification maze). Furthermore, the discrete block-based transaction processing of mainstream public chains is unsuitable for continuous agent interaction, and transaction fees for low-value transactions can be prohibitively high. All of these factors hinder the high-frequency, low-value micro-transactions of AI agents.

This is why Kite created a dedicated L1 blockchain network. It envisions AI agents as a new user category in the Web3 ecosystem, designed to support autonomous agents with programmable trust and AI-compatible capabilities. It integrates identity, payment, and behavior verification into a unified and composable protocol layer. By building a complete set of native economic infrastructure for intelligent agents, it enables agent-based commerce to operate securely and at scale.

The Kite team believes that in the future, the way people interact with the digital world will shift from direct human interaction to autonomous AI agents acting on their behalf. These agents will search for information, compare prices, place orders, sign contracts, manage subscriptions, and more, becoming the "new user interface." To achieve this, data must first be structured and verifiable. The next step is to build native identity, trust, and programmable payment mechanisms tailored for these agents.

Transforming from an analytics platform, it raises $33 million in funding to build an AI "dream team"

In fact, Kite didn't initially position itself as an infrastructure provider for autonomous agents. Kite, formerly known as Zettablock, positioned itself as an institutional-grade Web3 indexing and analytics platform, providing large-scale, real-time data support for networks like Sui, Polygon, Chainlink, and EigenLayer.

The rapid development of AI and the fact that the founding team members have experience and industry background in both blockchain and AI have given them the opportunity to transform into the Web3 AI track.

Kite's co-founder and CEO, Chi Zhang, holds a PhD in Machine Learning/AI (Statistics) and a Master's in Economics from the University of California, Berkeley. She previously led data engineering product development at Databricks and served as Chief AI Expert at dotData. Another co-founder, Scott Shi, who also serves as Kite's CTO, previously built real-time AI infrastructure at Uber and was an early engineer on Salesforce's Einstein AI team.

The two core members hold dozens of AI and blockchain-related patents and papers published at top conferences. The rest of the team also comes from companies like Uber, Databricks, Salesforce, and NEAR. With backgrounds from prestigious universities like Stanford, MIT, and the University of Tokyo, they possess extensive experience in blockchain protocol engineering and big data systems.

Earlier this month, Kite announced the completion of an $18 million Series A funding round led by PayPal Ventures and General Catalyst, with participation from 8VC, Samsung Next, SBI US Gateway Fund, Temasek's venture capital arm Vertex Ventures, Hashed, HashKey, Avalanche Foundation, LayerZero, and Animoca Brands. This round brings Kite's total funding to $33 million. The funds will be used to expand its agent trading platform and enhance the ability of AI agents to conduct large-scale micropayments using stablecoins on-chain.

PayPal Ventures has described Kite as "the first infrastructure purpose-built for the agent economy," noting that stablecoins and millisecond settlements are key technological gaps in AI agent systems, and that Kite provides a crucial bridge to these gaps. Furthermore, Kite is currently in a pilot phase, partnering with platforms like PayPal and Shopify to enable merchants to access the agent system through Kite's Agent App Store.

Modular architecture and Kite AIR

Kite's technical architecture is highly modular, focused on meeting the needs of AI agents. Its foundation is an EVM-compatible Layer-1 chain. Kite's official website currently advertises performance as "average block generation time of 1 second and near-zero fees."

The network's underlying operating environment is a customized KiteVM, and it utilizes a novel consensus mechanism called Proof of Attributed Intelligence (PoAI). PoAI combines proof-of-stake (PoS) with an attribution mechanism, enabling transparent attribution and rewards for model and data contributions to tasks performed by nodes while validating blocks. This means that every agent's task, including model invocation, data provision, and transaction completion, leaves an auditable record on-chain, ensuring fair rewards for all parties.

As infrastructure designed for large-scale, high-frequency AI agents, Kite's architecture prioritizes speed and scalability. Its cornerstone is a state channel mechanism that enables off-chain streaming micropayments and inter-agent communication with near-instant finality. Frequently transacting agents can open secure channels, enabling peer-to-peer, real-time micropayments or data exchanges without waiting for block confirmations. Billions of micro-events can be processed off-chain and periodically aggregated and settled on the main chain, significantly increasing throughput and reducing costs. This enables Kite to support streaming micro-transactions based on API calls, compute time, or data bytes, meeting the high-frequency billing requirements of the agent economy.

The Kite team has also launched a series of tools and modules for developers and agents. The platform's Kite AIR (Agent Identity Resolution) system is designed to provide agents with secure identity, policy enforcement, a verifiable system of record, and programmable payments executed on Kite's custom AI-native blockchain. Kite AIR's core components include KitePass for verifiable identity and policy enforcement, the Kite Agent App Store for marketplace and service discovery, and the Kite SDK & MCP Server for agent integration.

KitePass is Kite's agent identity module: each agent, dataset, or AI model can have a unique cryptographic identity, associated with corresponding permissions and reputation information. This identity system allows agents to be used across different services without repeated registration, while their operation history and permission scopes are tracked on-chain. Identity-based programmable governance allows agents to have fine-grained, automated permission control, such as setting limits on task types and fund usage, ensuring compliance with pre-defined rules at runtime.

The Kite Agent App Store is a unified marketplace and service discovery engine for service providers and autonomous agents. Service providers can list their products and monetize their APIs, AI models, data services, or business logic through automated payment processing, while gaining market access, identity-based trust, and usage analytics. For agents and developers, the App Store provides a direct service discovery channel, automatic settlement via the Kite settlement channel (every transaction is verifiable on-chain), complete usage history tracking, and an interoperable consumer workflow that connects identity, payment, and discovery.

Kite SDK and MCP Server are tool chains that connect applications to Kite's identity and settlement infrastructure: Kite SDK is aimed at agent developers, providing tools for building agents with verifiable identity, policy execution and on-chain settlement capabilities. It is suitable for creating autonomous agents, agent-driven business applications, cross-platform agent processes and prototype verification; MCP Server (Model Context Protocol server) is aimed at existing AI applications, enabling any MCP-compatible application to use Kite's identity and settlement functions, thereby allowing existing chatbots or AI assistants to participate in agent commerce, opening the door to agent capabilities for non-technical users, and realizing a bridge between traditional AI tools and the machine-to-machine economy.

Aero public beta to Ozone upgrade, hundreds of millions of calls, tens of millions of users

In February 2025, Kite launched its first public testnet, v1, codenamed Aero, on the Avalanche network. The network aims to enhance scalability and data processing capabilities while providing centralized coordination for AI workflows, including data providers, model builders, and autonomous agents. At the end of March, official statistics for the v1 Aero testnet were released, claiming that since its launch, the network has processed over 546 million AI agent calls, an average of approximately 11.4 million per day, executed approximately 32 million transactions, and connected approximately 4 million users, of which approximately 2.4 million are independent AI agent users.

After a first phase of exploration, in late May of this year, Kite AI upgraded its testnet, Aero, to Ozone, positioning it as an interactive portal for Agentic AI. The product narrative shifted from "scalable AI infrastructure" to "the foundational layer supporting the operation of the agent economy." The launch of Ozone further expanded the Kite AI ecosystem. According to Dune data , as of September 5th, the network had processed over 634 million AI agent calls and connected approximately 13.6 million users. Daily active accounts and new additions have remained at a high level since mid-August, with an average of 4 million daily active accounts.

In its official announcement of its Series A funding round, Kite began by stating its mission to “build the foundational layer for the Internet of Agents” and that its foundational layer powers the entire agent ecosystem through three pillars:

• Provide cryptographic identities for AI models, agents, datasets, and any digital service. Each AI “actor” or “asset” can maintain a unique and verifiable identity to support traceability, provenance, and governance.
• Programmable and fine-grained governance of delegated permissions, usage limits, and spending behavior – managing how AI agents operate autonomously “in the wild.”
• Instant proxy payments with near-zero fees enable autonomous systems to discover, negotiate, and pay for services with native access to stablecoins.

Steve Everett, Head of Global Market Development for Cryptocurrency and Digital Assets at lead investor PayPal, commented on the product, saying that its simultaneous atomic settlement via smart contracts, coupled with real-time tracking and auditing across high-performance blockchain protocols, is a killer combination for programmable payments in AI-powered commerce. This opens the door to a truly global, automated economy where people, businesses, and machines can interact easily and trustfully.

In summary, Kite's business model is deepening with the development of the intelligent agent economy. Its challenges lie in ecosystem development and technological iteration, while its strength lies in its early market presence. Whether it can stand out among numerous AI blockchain projects in the future depends on whether it can truly resolve the challenges of trust and settlement between intelligent agents, thereby providing a reliable foundation for automated economic activities.