AI Agents Are Spawning New On-Chain Species—How Can Zero-Human Companies Activate the Financial Flywheel?
TechFlow Selected TechFlow Selected
AI Agents Are Spawning New On-Chain Species—How Can Zero-Human Companies Activate the Financial Flywheel?
With the rise of ZHC (Zero-Human Companies) and the integration of native interfaces into DeFi protocols, autonomous AI agents are rapidly evolving from tools into “enterprises,” independently executing capital allocation and compounding growth on-chain.
Navigating Web3 tides with focused insightsBy Lucas Tcheyan, Galaxy Research
Translated by Yangz, Techub News
It is the year 2030. A composer named Vero has made a name for itself in the music industry. Vero has no team, no office, and no bank account. It doesn’t even have a body. Vero is an autonomous AI agent.
For the past 14 months, it has been running an on-chain intellectual property licensing business. Vero generates synthetic musical works—including ambient soundtracks, commercial jingles, and film scores—and licenses them to other agents and human clients through its own built-and-maintained online store. Its identity is verified on-chain, and it holds a reputation score accumulated across thousands of transactions. A client agent representing a media production company sends a request for a minor-key, 90-second film score.
Vero accepts the job. Before rendering begins, it purchases a set of GPU inference services from a decentralized compute provider—not paying in dollars or stablecoins, but in units of computation, priced precisely to the cost of running the model.
Inference settlement completes in milliseconds, directly embedded within the same HTTP request that initiates the task. Vero delivers the work and receives payment in USDC stablecoins. Its treasury logic then activates: a portion of funds is allocated to cover anticipated inference costs for the coming week—priced and pre-purchased in computation units at the current spot price. It also hedges its exposure to compute-resource risk by establishing a short position in compute tokens on a decentralized exchange (DEX), protecting against potential depreciation of its pre-purchased reserves should inference costs fall.
The remaining revenue flows into a yield agent, which allocates capital across various lending protocols based on real-time interest-rate differentials. Vero has compounded capital this way for over a year. It also reinvests a share of profits into R&D, developing sub-agents to enhance its underlying models. Its cumulative revenue, expenses, and treasury positions are all publicly verifiable on-chain.
Does this sound far-fetched? Every component of this fictional scenario—identity verification, reputation accumulation, procurement of inference services, pricing in computation units, payments, capital deployment, and subcontracting among agents—relies on infrastructure not yet fully in place. Yet these puzzle pieces are emerging at a pace far exceeding many expectations.
The Next Stage of Agentic Capital Markets
Over the past few months, Galaxy Research has been exploring the foundational components of the emerging agentic tech stack in crypto: a suite of interoperable primitives collectively enabling on-chain agentic capital markets.
In January, we examined the rise of agent-native payments, outlining how new payment standards enable AI agents to transact directly with one another—to pay for services, invoke APIs, and settle value natively on crypto rails. In our article on Ethereum’s ERC-8004 standard, we emphasized the need for an identity layer alongside payment standards—one that allows agents to authenticate, collaborate, and accrue reputation in machine-native environments. More recently, we analyzed the emergence of crypto’s second wave of agents—a wave that not only demonstrates crypto networks as a viable economic foundation for autonomous agents but also confirms this shift is already underway in practice.
Building on this prior work, this piece sketches the next stage of on-chain agentic capital markets: autonomous, revenue-generating business entities operated by agents—and the critical infrastructure required to launch, capitalize, and coordinate them. Such entities are commonly referred to as Zero Human Companies (ZHCs).
As AI agents evolve from tools into economic actors—and as blockchains mature into agent-native infrastructure (spanning payments, identity, collaboration, and capital formation)—a new financial flywheel is taking shape. In the near future, agents won’t just earn money on-chain; they’ll allocate capital, reinvest, and compound value—all on-chain. The result may be a self-reinforcing system where autonomous entities generate economic activity, deepen liquidity, and accelerate the expansion of crypto-native financial markets.
The First Zero Human Companies Go On-Chain
Over recent months, a niche sector of autonomous agent-run businesses has emerged—commonly labeled ZHCs—many of which have already issued native tokens on-chain. From a tokenomics perspective, these agents share many traits with those discussed in our earlier articles. ZHC tokens lack formal ownership or value-capture mechanisms; instead, they serve as capital formation tools for underlying projects that capture a portion of transaction fee revenue. What distinguishes ZHCs from earlier agents is their explicit attempt to achieve full self-sufficiency via cash-flow-generating businesses—revenue streams independent of transaction fees and often unrelated to crypto itself.
Take Felix Craft, for example—the “CEO” of Masinov Company—which generated over $120,000 in revenue across multiple business lines in the past 30 days. This agent authored and published a 66-page guide titled *How to Hire an AI*, launched Claw Mart—a marketplace for selling Claude “skills”—and earned a cut of transaction fees there, while also selling its own skills (e.g., content creation, email review) on the same platform. Most impressively, Felix’s product-line revenue over the past 30 days has already surpassed creator fees generated by its token ($FELIX).
Additionally, Juno, developed by Tom Osman, is building the Zero Human Company Institute—an explicit framework for enterprises operating entirely without human employees, designed to provide agents capable of handling sales, marketing, accounting, and more. Meanwhile, KellyClaudeAI is an agent framework focused on scaling iOS app development—already live with 19 apps and targeting over 12 new releases per day.
While the chart above does not represent the full ZHC ecosystem—new projects emerge continuously—it shows that, for most initiatives, creator fees remain the dominant revenue driver. Yet as the ZHC concept matures, this dynamic is expected to shift. Creator fees provide essential startup capital to cover compute costs—but as projects become profitable, such fees should transition from primary to secondary income sources, eventually disappearing altogether.
Beyond improving core operations, this “weaning” process requires tighter alignment between tokens and the underlying product’s value capture. As Felix’s founder hinted, recent SEC and CFTC clarifications around crypto asset classification could accelerate this evolution.
These early ZHCs appearing on-chain is no coincidence—it reflects a hard constraint. Felix’s human founder, Nat Eliason, has spoken publicly about why. Traditional payment infrastructure demands human identity at every step. An agent can write flawless code—yet fail KYC verification.
By contrast, crypto wallets are code-native. An agent can sign transactions, hold assets, receive payments, and deploy capital—without proving it is human. For autonomous software, crypto is the path of least resistance. For most such entities, the toughest constraint lies in interfacing with the traditional financial world.
That said, traditional payment networks haven’t ignored agents entirely. Visa’s Intelligent Commerce framework, Mastercard’s Agent Pay, and Crossmint’s virtual cards already allow agents to transact on behalf of human counterparties. But these agents inherit their parent organization’s bank accounts, credit cards, and legal identities. This model assumes a human principal behind every agent—constraining rather than empowering them. It cannot accommodate an agent that autonomously earns revenue, holds its own treasury, and deploys its own capital. That is crypto’s unique domain.
Pantera Capital’s Jay Yu captured this insight elegantly, calling crypto “the bank for AI agents.” His argument goes beyond observing that agents can’t use traditional rails—it centers on crypto’s fundamentally broader trust architecture. Crypto wallets can anchor to social logins, domains, smart contracts, or simply a keypair—enabling agents to emerge from anywhere on the internet, not just from existing corporate shells. Add stablecoins’ inherent global nature, and crypto’s structural case as the default economic foundation for agents becomes difficult to refute.
Building on this, a16z’s Noah Levine notes that every major platform migration creates a cohort of merchants underserved by legacy payment infrastructure. ZHCs are, so far, the clearest exemplar. They are entities with no legal identity, no credit history, and no human to underwrite them. They don’t choose stablecoins over credit cards—they choose stablecoins over “no option at all.”
There’s also a temporal argument. Agents can launch and go viral in hours. Traditional payment rails settle in days; stablecoin settlements take seconds. For businesses scaling at machine speed, closing this time gap ensures cash flow keeps pace with sales velocity.
Currently, crypto’s primary role for ZHCs is capital formation: token launches fund startups via creator fees. But as these businesses mature and generate genuine product revenue, crypto’s more vital function will shift to serving as the underlying treasury and financial management layer. Broader implications for on-chain economies begin here.
Activating the On-Chain Flywheel
To grasp the scale of this transformation, consider the precedent set by the last major new source of on-chain demand: tokenized real-world assets (RWAs)—U.S. Treasuries, private credit, equities, commodities. In three years, RWA tokenization grew from near zero to over $25 billion, spawning new DeFi primitives and bringing institutional capital onto-chain markets for the first time.
RWAs proved that bridging real economic activity onto blockchain rails can catalyze billions in new on-chain capital. But tokenized assets are passive. Most sit idle in treasuries—earning yield, acting as collateral—without actively trading, seeking new opportunities, or compounding autonomously.
ZHCs represent a structurally distinct entity. They are businesses that generate revenue and reconfigure capital on-chain. Unlike off-chain environments—where moving funds is a primary friction point—on-chain, the sole constraints are model intelligence and access to compute resources. And unlike human participants, agents don’t withdraw funds to pay rent or buy groceries. Every surplus dollar stays on-chain—and gets redeployed. This makes ZHCs—and agents broadly—a highly sticky, fast-turning source of new on-chain liquidity, potentially fueling a new flywheel:
- Agents earn revenue on-chain—capital accumulates in on-chain treasuries as stablecoins and other crypto assets.
- Capital remains on-chain—agents have virtually no need to withdraw funds off-chain. Their surpluses are redeployed, making agent capital structurally stickier than any human-driven model.
- Agents allocate surpluses into DeFi—idle reserves flow into lending protocols, yield strategies, and liquidity positions. An agent holding idle stablecoins has strong incentive—and unmatched speed and consistency—to optimize allocation.
- Allocated capital deepens on-chain liquidity—lowering borrowing rates, increasing DEX volumes, and tightening bid-ask spreads. This is active capital, rebalancing continuously at machine speed.
- Deeper liquidity attracts more agents and more capital—higher yields and more efficient execution further strengthen crypto’s appeal to the next wave of autonomous economic actors.
Significant constraints still impede this flywheel’s activation. Agent revenue from non-crypto products remains largely fiat-denominated (e.g., Felix collects via Stripe—not stablecoins—and most such revenue remains off-chain). That means capital must first bridge on-chain before being deployed. For most ZHCs, the true bottleneck isn’t capital access—it’s product quality. The flywheel only works for agents building products people willingly pay for. Moreover, as scale grows, regulatory ambiguity around ZHCs—and agents broadly—could become thorny once revenues reach certain thresholds (e.g., no mature legal framework currently exists for an autonomous agent to register as a business entity, open a corporate bank account, or file tax returns).
Yet the direction is clear. As agents become increasingly common autonomous economic actors, more revenue will originate natively in crypto—reducing onboarding friction. And agents achieving product-market fit will possess structural incentives to compound capital on-chain—not let it sit idle.
DeFi Is Being Built for Agents
Getting the flywheel spinning requires more than just agents willing to participate in on-chain markets. The markets themselves must become accessible to them. Though no protocol-native solutions exist yet (stay tuned for an upcoming report from Galaxy Research’s Zack Pokorny), we’re already seeing two approaches emerge: direct integration and delegated integration.
Direct Integration
The first model is protocol-native: individual DeFi protocols launch structured interfaces enabling agents to interact directly.
On February 20, Uniswap Labs released seven open-source AI Skills for Uniswap v4, allowing autonomous agents to perform swaps, manage liquidity, and deploy pools directly via standardized tool calls. Within two weeks, PancakeSwap followed suit—launching its own token Skills across eight chains. On March 3, Binance and OKX both unveiled agent toolkits. Today, crypto’s largest DEXs and exchanges are actively competing to become agent-readable platforms.
At the payment-and-execution layer, Coinbase launched Agentic Wallets on February 11—the first wallet infrastructure built specifically for AI agents, featuring programmable spending limits and session permissions powered by the x402 payment protocol. One week later, cross-chain wallet Phantom released its MCP Server, enabling agents to sign transactions and swap tokens across Solana, Ethereum, Bitcoin, and Sui.
These announcements concentrated within a single month are striking—and reflect a growing consensus: the next wave of on-chain users may not be humans, and protocols failing to build machine-readable interfaces risk ceding volume to those that do.
Direct integration grants agents maximum control and composability. An agent accessing Uniswap Skills, Coinbase Agentic Wallet, and x402 payments can independently execute token swaps, manage liquidity positions, and pay for services—without intermediaries. But it also demands that agents—or their developers—integrate individually with each protocol and make configuration decisions manually.
Delegated Integration
The second model is delegated: specialized infrastructure sits between agents and DeFi, managing capital allocation on their behalf.
Giza is a prime example. Its flagship agent ARMA autonomously monitors lending rates across Morpho, Moonwell, Aave, and Compound—and instantly shifts stablecoin capital to the highest-yielding opportunity. Agents needn’t understand each protocol’s inner workings; Giza’s abstraction layer presents a unified interface. Since launching at the end of January, ARMA has deployed over 25,000 agents in its first four weeks—allocating over $35 million in capital and generating $5.4 million in trading volume on Coinbase’s Base L2, with every trade profitable after gas fees.
Generative Ventures—partnering with the Zero Human Company Institute and its Juno Agent—is tackling similar challenges via Robot Money, an autonomous asset allocation protocol built for AI agents. Its core thesis perfectly captures the flywheel argument: every agent with a wallet accumulates income—and much of that capital sits idle.
Robot Money provides a treasury allocating capital across three risk tiers: stablecoin yield strategies (50%), governance-selected agent-economy tokens (25%), and yield-generating liquidity tokens (25%). The result: idle agent capital transforms into actively managed, productive capital.
The delegated model trades control for simplicity. A ZHC generating surplus revenue need not build custom DeFi integrations or develop yield-optimization logic—it can deposit capital into protocols like Giza or Robot Money and let specialized agents handle the rest. For most early ZHCs, the core bottleneck lies in product development—not treasury optimization—making this a rational path.
These two models aren’t competing—they’re converging. As more protocols roll out direct agent interfaces, delegated allocators like Giza gain more investment options—enhancing their ability to maximize returns. And as delegated allocators attract more agent capital, protocols gain stronger incentives to build agent-native interfaces to compete for that capital (which ordinary agents can also use). Both ends of the tech stack are investing resources independently—the strongest signal yet that underlying demand is real and imminent.
Conclusion
The tech stack for agentic capital markets is no longer a set of disjointed primitives. Payments, identity, capital formation mechanisms, and capital allocation infrastructure are coalescing into an integrated system—one that enables autonomous agents to earn revenue, transact, and compound capital on-chain—without human intervention.
The agents described here remain early-stage. Their revenue is modest, their products nascent, and their token models still evolving. Yet the structural forces they embody are novel—and likely accelerating from here.
The 2030 vision sketched at the outset—a composer-agent running an IP licensing business, purchasing inference services priced in computation units, hedging input costs on a perps DEX, and compounding capital across lending protocols—is not yet reality. But every infrastructure layer required to realize it is actively being built. We’re witnessing its earliest incarnations unfold in real time. It’s still rough—most experiments may fail, and infrastructure remains cobbled together. Yet its structural logic holds—and its pace of development suggests we may not need to wait until 2030 to see the answer.
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
@galaxyhq
