When machines start paying: Kite AI and the settlement infrastructure of agentic payments
TechFlow Selected TechFlow Selected
When machines start paying: Kite AI and the settlement infrastructure of agentic payments
How big is the market for the Agentic Internet? PayPal is willing to spend big.
Navigating Web3 tides with focused insightsWritten by: ChandlerZ, Foresight News
The first act of generative AI showed us intelligent tools that can chat, write, and draw. But what truly determines whether it can enter the mainstream economy isn't how human-like its responses are, but whether it can independently complete a real transaction—identifying counterparties, verifying identities and permissions, making automatic payments, recording fulfillment, triggering arbitration or refunds when necessary, and more. Traditional internet payment and risk control assume humans are in the loop, with SMS verification, manual approval, and customer service as the norm. Yet for high-speed machine-to-machine collaboration, micro-payments at AI Agent frequency, pay-per-call billing, clear revenue sharing, and accountability require a more foundational, consistently verifiable public framework.
Addressing this gap, AI payments have become central to the next wave of infrastructure narratives—moving identity, permissions, payments, governance, and attribution from temporary application-layer agreements into a settlement and coordination base that anyone can verify. It is within this context that Kite AI emerges, earning recognition from PayPal and numerous Wall Street institutions. Rather than simply adding another payment SDK on top of existing blockchains, Kite AI attempts to integrate Agent passports, native settlements, programmable governance, and transparent attribution into a unified execution environment at the chain level, enabling AI Agents to transact with standardized capabilities.
Can crypto + AI achieve true mass adoption? How large could the market for next-generation crypto + AI payment infrastructure be? What will future payment infrastructure look like? The answer is simple: without verifiable payment and clearing mechanisms, there can be no scalable AI Agent economy. Using this metric, we examine Kite AI to understand what it is and what problems it aims to solve.
The Watershed Moment for AI Is Payment
In recent years, the generative capabilities of AI Agents have steadily expanded. But to truly deploy them in production systems, action—not expression—is key. Tasks such as automatic ordering, price comparison, invoicing, tracking, and after-sales support become challenging when linked to fund disbursement, because they demand requirements unmet by traditional payment gateways: machine-frequency authorization, programmable budgets and whitelists, cross-platform portable exception handling, and publicly verifiable value attribution. Enterprise demands for AI have shifted from "writing reports" to "running closed loops," especially evident in high-frequency scenarios like e-commerce, subscriptions, transportation, SaaS, and data services.
Leaving rules at the application or gateway layer leads to fragmentation. Each merchant integration or new scenario requires rewriting permission, budget, risk control, and after-sales processes, causing collaboration costs to rise nonlinearly with scale. A better approach is to move verifiable identity, native settlement, programmable governance, and public attribution down to the chain level, making them default public goods. This way, AI Agents carry verifiable capabilities rather than stacks of ad hoc agreements; counterparties only need to validate passports and policies before accepting tasks; developers benefit from unified encapsulation, significantly reducing migration complexity.
Kite AI has chosen this bottom-up path. With deep expertise in data and distributed systems, the team’s engineering philosophy is to embed complexity into the protocol once and for all, rather than repeatedly patching applications. This explains their shift from data infrastructure to one of transaction and trust—building the “money and rules” essential for AI Agents into foundational layers so machine-to-machine transactions have enforceable foundations.
Kite AI is building the foundational transaction layer for an “Agentic Internet,” providing unified identity, payment, and governance infrastructure for autonomous agents. On February 10, 2025, Kite AI launched the testnet of PoAI (Proof of AI), its L1 sovereign blockchain. According to official data, within less than 70 hours of v1 going live, the testnet surpassed 230,000 wallet connections and exceeded 2.5 million AI inference calls.
In September 2025, Kite AI announced $33 million in total funding, including an $18 million Series A round led by PayPal Ventures and General Catalyst. Funds will accelerate development of the Agentic Internet. Integration with Shopify and PayPal is now live—any PayPal or Shopify merchant can join via the Kite Agent App Store and become discoverable by AI shopping agents. All transactions settle on-chain with full traceability, using stablecoins and programmable permissions.
An L1 Built for Agents: How to Make Machines Pay
To move AI Agents beyond demos into real business operations, payment and trust must be system defaults, not add-ons. This is why Kite AI chose to rebuild from the ground up—consolidating who can do what, how payments happen, error handling, and value distribution into a single execution environment. Only when identity, rules, and funds are bound by the same deterministic constraints on one ledger will Agent authorization boundaries avoid reinterpretation across intermediate layers, and transaction evidence remain reusable across contexts and counterparties. Thus, Kite AI follows an EVM-compatible sovereign chain path, anchoring core primitives of trust and payment at L1 through Kite AIR (Agent Identity & Resolution).
Agentic Payment Layer: Making Spending Power and Accountability System Defaults
In this architecture, KitePass is the first cornerstone. It's not just an image of an ID card, but an executable on-chain passport. It specifies the Agent’s creator, authorized capabilities, budget caps and whitelists, valid time windows, and conditions requiring multi-signature or escalation to human review. Counterparties don’t need custom risk checks—they simply verify compliance of the passport and rules to approve transactions.
For Agents, minimal privilege is the default state, reducing systemic risk. Addressing frequent payment needs, Kite AI makes micropayments a ledger default. Using state channels and dedicated payment channels, it handles high-volume, low-value deduction requests with millisecond confirmations and low fees. Pay-per-call, micro-subscriptions, and streaming payments thus become standard, while only channel opening, final settlement, and dispute resolution return to the main chain—balancing speed and accountability.
To ensure rules aren't just paper promises, Kite AI introduces session keys and programmable spending strategies, embedding rate limits, quota circuit breakers, automatic refunds, freeze/unfreeze actions, and dispute escalation into a unified governance state machine. Paired with on-chain auditing and reputation systems, compliance and violations accumulate into long-term, referenceable credit. Together, these functions equate to a corporate card + risk console + reconciliation ledger at the ledger layer—offering standardized spending proofs and auditable trails externally, and callable primitives and policy templates internally for developers.
Transparent Attribution and Fair Compensation: Recording Who Did the Work in One Ledger
Another challenge in the AI Agent economy is determining who contributed and how rewards should be split. Behind a seemingly simple transaction often lies a symphony of model inference, data API calls, tool execution, and multi-Agent orchestration. Without credible attribution, revenue sharing either relies on platform decisions or off-chain settlements, failing to create sustainable incentives. Kite AI addresses this by forming a closed loop across evidence, measurement, distribution, and dispute resolution:
-
Evidence: Every call and output leaves a verifiable execution record specifying who did what, when, and under which rules
-
Measurement: Metrics map usage by call count, duration, quality, or outcome-based indicators
-
Distribution: Revenue-sharing logic is encoded in contracts, automatically splitting payouts among model providers, data sources, tool operators, and orchestrators
-
Dispute: Quality or attribution disputes enter the governance state machine, triggering re-evaluation, rollbacks, and redistribution—all recorded on-chain
With this closed loop, participants are willing to offer their capabilities to AI Agents because evidence is accessible, revenue splits are predictable, and dispute paths exist—no longer dependent on platform rulings.
Ultimately, underlying capabilities must be productized to gain adoption. Kite AI offers an Agent App Store for merchants and service providers—a hub currently listing 36 services including Shopify, Uber, Masa, and Irys. Any team can publish an Agent for direct user access; merchants list capabilities, pricing, and SLAs; Agents follow a standard path of discovery, invocation, settlement, and traceability.
Compared to other AI × blockchain approaches in the market, Kite AI deliberately differentiates itself. General-purpose L2s tend to accommodate AI Agents via account abstraction and plugins; compute networks focus on decentralized orchestration of inference tasks; vertical app chains target industry-specific closed loops. Kite AI places AI Agent identity and payment at the system core, closing governance and attribution within the same ledger as transactions. It doesn't aim to replace all components, but fills the critical missing piece—how machines can spend money trustlessly—becoming a settlement and coordination layer others can rely on.
Imagine a shopping AI Agent placing a coffee order within your $50 budget, constrained by its passport. The merchant verifies the passport and approves. Payment locks in a channel with an SLA for refund if delivery is late. Delivery status updates are written to the chain per rules; anomalies trigger circuit breaking or refunds. Upon completion, the contract automatically releases payment and tips, splitting revenue among data, model, and tool providers based on the evidence trail. Performance feeds into the reputations of both merchant and Agent. No customer service needed—economic activity is autonomously completed under clear constraints and transparent audits. This is exactly how Kite AI, as an L1 built for AI Agents, turns “machines paying” into everyday reality.
x402 as Entry Point, Kite Lays the Settlement Rails for Agent Payments
Recently, x402 has generated market excitement, and Kite AI serves as the optimal L1 execution and settlement layer for the x402 protocol and emerging Agentic economy. x402 is a natively internet-based payment standard proposed by Coinbase, leveraging the long-unused HTTP 402 (Payment Required) status code to allow web pages and APIs to express “payment required” directly at the protocol layer. Naturally supporting stablecoins, small frequent payments, and low latency, x402 enables AI Agents to instantly purchase data, tools, and compute power per call, while allowing Web2 services to integrate on-chain settlement with minimal changes—bypassing the complex hurdles of registration, email, OAuth, or intricate signatures.
Looking deeper through this entry point, the relationship between x402 and Kite AI becomes clear. x402 clearly signals “payment required” in the HTTP world and provides an authorization format, but does not handle who pays, how much they’re allowed to spend, how over-authorization is blocked, how failed payments are refunded, how SLA breaches are arbitrated, or how multi-party revenue is split. These runtime settlement and governance issues are precisely what Kite AI solves at the chain level. Designed from the start to natively align with x402’s payment primitives, Kite AI is among the first fully compatible L1s, enabling AI Agents to initiate, receive, and reconcile payments via standardized intent. As more AI Agent-focused payment standards emerge, the combination of “entry via x402, settlement via Kite AI” will generate stronger network effects. Recently, Kite AI announced a strategic investment from Coinbase Ventures, extending its previously announced $33 million financing round led by PayPal Ventures and General Catalyst. New funds will support development of autonomous AI Agent payment infrastructure and joint efforts with Coinbase to accelerate widespread adoption of the x402 protocol.
Only by strengthening the broader ecosystem can the full picture emerge. Kite AI lowers integration barriers through App Stores, SDKs, and templates, while modular offerings close the value chain loop: For example, Codatta, incubated by Kite AI, packages high-quality data as a verifiable service, charging by usage and distributing rewards by contribution—naturally benefiting from x402’s standardized payment entry and Kite’s settlement and distribution capabilities. Codatta previously received joint funding from Kite AI and the Avalanche Foundation through the InfraBUIDL (AI) program, a $15 million initiative. Meanwhile, collaboration with Brevis embeds verifiable computing and privacy-preserving audits into payment and fulfillment workflows, ensuring every step of evidence, measurement, distribution, and dispute resolution is on-chain and auditable—reducing enterprise concerns around compliance and risk when adopting AI Agent payments. Together, x402 ensures the message “payment required” is correctly and smoothly conveyed across the internet, while Kite ensures money is paid securely, compliantly, and accountably, with resulting value transparently settled. Their synergy forms the standard swipe method for the AI Agent era.
Funding, Community, and Future Potential
Kite AI has raised $33 million in total funding to date, with its latest Series A round amounting to $18 million, led by PayPal Ventures and General Catalyst, and joined by 8VC, Samsung Next, SBI US Gateway Fund, Vertex Ventures, Hashed, HashKey Capital, Dispersion Capital, Alumni Ventures, Avalanche Foundation, GSR Markets, LayerZero, Animoca Brands, Essence VC, and Alchemy.
In addition, Kite AI recently secured further strategic investment from Coinbase Ventures, continuing its $33 million round led by PayPal Ventures and General Catalyst.
For a network aiming to support Agent payments, this shareholder structure sends several signals:
-
First, resources and brand strength from the payment ecosystem boost merchant onboarding, regulatory dialogue, and pilot partnerships;
-
Second, leading dollar funds signal sustained capacity for long-term engineering and business development;
-
Third, participation by tech and industry capital enhances connectivity with upstream models and downstream applications.
This is also the origin of the market label “PayPal’s heir”—not implying ownership, but highlighting natural affinity with payment systems.
Usability is validated by testnet scale metrics. Since 2025, the network has reached over 17 million users, with approximately 1.7 billion cumulative interactions. Daily active users consistently range between 2 and 4 million, reflecting a usage pattern characterized by high frequency, low value, and auditability. These figures are not isolated peaks, but reflect stability under real traffic for the combined mechanism of per-call deductions, programmable governance, and channel-based settlement. In high-concurrency inference and payment scenarios, channels handle vast volumes of micro-transactions, the governance state machine executes anomaly responses without human intervention, and audit trails provide evidence for subsequent reviews and revenue splits.
On the roadmap, Kite AI plans to launch its mainnet in Q4 2025, delivering core features including native stablecoin payments, refined Agent ID and trust layers, and continued expansion of the App Store and SDK ecosystem to deepen the supply side of agent-as-service providers. Its capital structure provides ample momentum. For an L1 centered on payment and trust, this investor mix strengthens anticipated synergies with the payment ecosystem and offers sufficient flexibility for engineering and commercial investments during the mainnet push.
The Battle for the Clearing Layer Rests on “Letting Agents Spend Safely”
As AI evolves from talking to acting, payment and clearing cease to be peripheral interfaces and become the system itself. Kite AI’s solution consolidates passport (identity/authorization)—payment (native stablecoin)—governance (risk control/arbitration)—attribution (transparent revenue sharing) into chain-level public goods. Its challenge lies in turning this suite of capabilities into an engineering reality: reusable across scenarios, with predictable cost and finality, and interoperable with existing ecosystems.
If e-commerce loops run stably, with key indicators trending positively across supply, demand, and health metrics, further enhanced by mainnet interoperability and regulatory stability, then “machines starting to pay” will no longer be a demo, but a default capability of the next-generation internet. At that point, Kite AI will be one of the clearing rails for the Agent economy—working invisibly to pay on your behalf, reconcile accounts, manage risks, and distribute value fairly for you and your Agents.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
Foresight News
