Written by: Lucas Tcheyan, Galaxy Research
Compiled by: Yangz, Techub News
The year is 2030. A composer named Vero has made a name for himself in the music industry. Vero has no team, no office, and no bank account. He doesn't even have a body. Vero is an autonomous AI agent.
For the past 14 months, it has been operating an on-chain intellectual property licensing business. Vero generates synthesized music, including ambient soundtracks, commercial jingles, and movie scores, and licenses them to other agents and human clients through an online store it has built and maintains. Its identity is verified on-chain and it has a reputation score accumulated over thousands of transactions. A client agent representing a media production company sent a request for a minor key, a 90-second movie score. Vero took the job, and before starting rendering, it purchased a set of GPU inference services from a decentralized computing service provider, paying not in dollars or stablecoins, but in units of computation, with the transaction price accurate to the cost of running the model.
Inference settlement is completed within milliseconds, directly embedded in the same HTTP request that initiates the task. Vero delivers its work, receives USDC stablecoin payment, and its treasury logic immediately activates. A portion of the funds is used to cover the expected inference costs for the following week, which are pre-purchased and valued in computing units at the current spot price. It also hedges its exposure to computing resources by establishing short positions in computing tokens on decentralized exchanges (DEXs) to prevent the pre-purchased reserves from devaluing due to declining inference costs. The remaining revenue is channeled to a yield agent, which allocates the funds to different lending protocols based on real-time interest rate differences. Vero has been compounding its capital in this way for over a year. It also reinvests a portion of its profits into R&D, developing sub-agents to enhance the underlying model. Its cumulative revenue, expenditure, and treasury position are all publicly visible on-chain.
Does this sound unbelievable? Every step in this fictional scenario—identity verification, reputation building, procurement of inference services, unit pricing, payments, capital deployment, and subcontracting between agents—requires infrastructure that is not yet fully in place. But these pieces of the puzzle are emerging at a rate far exceeding many people's expectations.
The next stage of the proxy capital market
Over the past few months, Galaxy Research has been exploring the underlying structure of an emerging proxy technology stack in the crypto space: a set of underlying components that collectively enable on-chain proxy capital markets.
In January of this year, we explored the rise of proxy payments, illustrating how new payment standards enable direct transactions between AI agents to pay for services, call APIs, and natively settle value on a crypto track. In our article on the Ethereum ERC-8004 standard, we emphasized the need for an identity layer alongside payment standards, allowing agents to authenticate, collaborate, and build reputation in a machine-native environment. More recently, we analyzed the emergence of a second wave of proxy activity in the crypto space, demonstrating not only that crypto networks provide a viable economic foundation for autonomous agents but also that this shift is already underway in practice.
Building on previous research, this article outlines the next phase of the on-chain proxy capital market: autonomous revenue-generating business entities operated by proxies, along with the critical infrastructure required to support their establishment, capitalization, and collaborative operation. These entities are often referred to as Zero Human Companies (ZHCs).
As AI agents evolve from tools into economic actors, and blockchain matures into a native infrastructure for agents (encompassing areas such as payments, identity, collaboration, and capital formation), a new financial flywheel is taking shape. In the near future, agents will not only be able to earn money on-chain, but also allocate capital, reinvest, and compound their value on-chain. The result could be a self-reinforcing system in which autonomous entities create economic activity, deepen liquidity, and accelerate the expansion of crypto-native financial markets.
The first batch of zero-human companies to land on the blockchain
In recent months, a niche industry comprised of autonomous proxy services, often referred to as ZHC, has emerged, many of which have already issued corresponding tokens on-chain. From a tokenomics perspective, these proxies share many characteristics with the proxies discussed in previous articles. ZHC tokens lack formal ownership or value capture mechanisms, instead serving as a capital formation tool for underlying projects that generate revenue from transaction fees. What distinguishes ZHC from earlier proxies is that they also attempt to achieve complete self-sufficiency through cash flow-generating businesses unrelated to transaction fee revenue and typically unrelated to the crypto space itself.
Take Felix Craft , for example. As the "CEO" of the Masinov Company, it has generated over $120,000 in revenue from multiple business lines in the past 30 days. This agent authored and published a 66-page guidebook, "How to Hire an AI," and launched a marketplace called Claw Mart to sell Claude "skills," earning a portion of the transaction fees. It also sells its own skills (such as content creation and email moderation) on the same marketplace. Most impressively, in the past 30 days, Felix's revenue from its product lines has exceeded the creator fees generated by its token ($FELIX).
In addition, Project Juno , developed by Tom Osman, is building a zero-human-employee company research institute. This is a clear framework for business entities that operate entirely without human employees, aiming to provide a set of agents capable of handling various tasks from sales and marketing to accounting. KellyClaudeAI , on the other hand, is an agent framework focused on scaling iOS app development, currently featuring 19 apps and aiming to launch more than 12 new products daily.
While the above chart doesn't represent the entire ZHC ecosystem (new projects are constantly emerging), it shows that creator fees remain the primary revenue driver for most projects. However, this landscape is expected to shift as the ZHC concept matures. Creator fees provide the computational cost funding needed to launch projects, but as projects become profitable, they should gradually transition to a secondary revenue source and eventually be phased out. Beyond improvements in the underlying business, this "weaning" process requires a better alignment between the token and the value capture of the underlying product. As hinted by the founders of Felix, the recent clarification of crypto asset classifications by the SEC and CFTC may accelerate this process.
The appearance of these early ZHC instances on-chain is not accidental, but rather a consequence of a real-world constraint. Nat Eliason, the human founder of Felix, has publicly discussed the reasons behind this. Traditional payment infrastructure requires human identity at every stage. An agent can write code fluently but cannot pass KYC verification. In contrast, crypto wallets are code-native. An agent can sign transactions, hold assets, receive payments, and deploy capital without proving their human identity. For autonomous software, crypto is the path of least resistance. For most of these entities, the most challenging limitation lies in the need to interact with the traditional financial world.
This is not to say that traditional payment networks ignore agents. Tools such as Visa's Intelligent Commerce framework, Mastercard's Agent Pay, and Crossmint's virtual cards already allow agents to conduct transactions on behalf of human counterparties. However, these agents inherit the bank accounts, credit cards, and legal identities of their parent organizations. This model assumes that each agent has a human principal behind them. They are constrained by this constraint, rather than empowered. It cannot accommodate an agent that independently earns income, holds its own treasury, and deploys its own capital. And this is precisely the unique application scenario for crypto.
Jay Yu of Pantera Capital articulates this succinctly, describing crypto as "the bank of AI agents." His argument goes beyond simply observing that agents cannot be accessed through traditional channels; it extends to how crypto underpins a fundamentally broader trust structure. Crypto wallets can be anchored to social logins, domains, smart contracts, or simply a key pair. This allows agents to emerge from anywhere on the internet, not just from existing corporate shells. Coupled with the inherently global nature of stablecoins, the structural argument for crypto as the default economic foundation of agents becomes difficult to refute.
Building on this, Noah Levine of a16z points out that every platform migration creates a new batch of merchants that existing payment infrastructure cannot serve. ZHC is the clearest example to date. These are entities with no legal person status, no credit history, and no human insurer to underwrite them. They didn't choose stablecoins over credit cards; they chose stablecoins over "nowhere else to go."
Furthermore, there's an argument based on time. Agents can launch a product and quickly gain popularity within hours. Settlement through traditional payment channels takes days, while stablecoin settlements take only seconds. For businesses scaling at machine speed, reducing this time lag allows cash flow to keep pace with sales.
Currently, the primary role of crypto technology for ZHC is capital formation. Token issuance provides initial funding through creator fees. However, as these businesses mature and generate real product revenue, crypto technology will play a more significant role as the underlying treasury and financial management layer. A broader impact on the on-chain economy is thus beginning to emerge.
Activate the flywheel on the chain
To understand the potential scale of this shift, it's helpful to look back at the precedent set by the last major source of new on-chain demand. The tokenization of real-world assets (US Treasury bonds, private credit, stocks, and commodities) grew from near zero to over $25 billion in three years, spawning new DeFi underlying components and bringing institutional capital into on-chain markets for the first time.
RWA demonstrated that bridging real-world economic activity onto the blockchain could catalyze billions of dollars in new on-chain capital. However, tokenized assets are passive. They mostly sit idle in vaults, earning yield and serving as collateral. They don't actively trade, seek new opportunities, or compound themselves.
ZHC represents a structurally distinct entity. They are enterprises capable of generating revenue and reallocating capital on-chain. Unlike off-chain environments where fund flows are the primary source of friction, on-chain, the only constraint lies in the intelligence of the model and its access to computational resources. Furthermore, unlike human participants, agents do not need to withdraw funds to pay rent or purchase necessities. Every surplus can remain on-chain and be used for reallocation. This makes ZHC, and agents more broadly, a sticky and rapidly circulating source of new on-chain liquidity, potentially spawning a new flywheel:
Agents earn income on-chain—this capital accumulates in an on-chain treasury in the form of stablecoins and other crypto assets.
This capital remains on-chain—agents have virtually no need to withdraw funds off-chain. Their surpluses can be reallocated, making agent capital structurally more sticky than any human-driven model.
Agents allocate surpluses to DeFi—idle reserves are directed to lending protocols, yield strategies, and liquidity positions. An agent holding idle stablecoins has a strong incentive to optimize allocations, and their speed and consistency are unmatched by any human.
The allocated capital deepens on-chain liquidity—which is expected to lower lending market rates, increase DEX trading volume, and narrow bid-ask spreads. This is active capital, continuously rebalancing at machine speed.
Deeper liquidity attracts more agents and more capital—higher yields and more efficient execution will further enhance the on-chain attractiveness to the next wave of autonomous economic actors.
Significant constraints remain hindering the launch of this flywheel. Agent revenue for non-crypto products still primarily derives from fiat currency (e.g., Felix receives payments through Stripe, not stablecoins, and most of this revenue remains off-chain), meaning capital must first be on-chain before it can be allocated. For most ZHCs, the real constraint isn't capital acquisition, but product quality. The flywheel only works for agents who can create products people are willing to pay for. Furthermore, as scale increases, the lack of regulatory clarity surrounding ZHCs (and agents more broadly) can lead to thorny issues once revenue reaches a certain level (e.g., there's currently no mature legal framework allowing an independent agent to register as a business entity, open a corporate bank account, or file tax returns for its income).
But the direction is clear. As agents become increasingly common autonomous economic entities, more revenue will be generated directly in the native form of cryptocurrency, and friction on the blockchain will decrease. Agents who successfully achieve product-market fit will have a structural incentive to realize compound interest on the blockchain, rather than letting their funds sit idle.
DeFi is being built for agents
For the flywheel to spin, it's not enough to simply get agents willing to participate in on-chain markets. The markets themselves must also become accessible to them. While there's no protocol-native solution yet (stay tuned for an upcoming report from Zack Pokorny of Galaxy Research), we're beginning to see two approaches addressing this: direct integration and delegated integration.
direct integration
The first model is protocol-native, where various DeFi protocols launch structured interfaces that allow agents to interact directly with them.
On February 20th, Uniswap Labs released seven open-source AI Skills for Uniswap v4, enabling autonomous brokers to directly perform exchanges, liquidity management, and pool deployment through standardized tool calls. Within two weeks, PancakeSwap followed suit, launching its own token Skills across eight chains. On March 3rd, Binance and OKX both released their broker toolkits. The largest DEXs and exchanges in the crypto space are now actively competing to become broker-readable platforms.
On the payment and execution front, Coinbase launched Agentic Wallets on February 11, touted as the first wallet infrastructure specifically designed for AI agents, featuring programmable spending limits and session permissions based on the x402 payment protocol. A week later, cross-chain wallet Phantom released its MCP Server, enabling agents to sign transactions and exchange tokens on the Solana, Ethereum, Bitcoin, and Sui networks.
These releases, concentrated within a single month, are remarkable. This also reflects a consensus: the next wave of on-chain users may not be human, and protocols that fail to build machine-readable interfaces may hand over transaction volume to those that do.
The direct integration model gives brokers maximum control and composability. A broker with access to Uniswap Skills, Coinbase Agentic Wallet, and x402 payments can independently perform token swaps, manage liquidity positions, and pay for services without intermediaries. However, this also requires the broker (or its developers) to integrate with each protocol individually and make their own configuration decisions.
Delegated Integration
The second model is the delegated model, which involves building dedicated infrastructure between the agent and DeFi to handle capital allocation on behalf of the agent.
Giza is a prime example. Its flagship agent, ARMA, autonomously monitors lending rates on protocols such as Morpho, Moonwell, Aave, and Compound, and transfers stablecoin funds to the highest-yielding opportunities in real time. Agents don't need to understand the specific operations of each protocol; Giza's abstraction layer translates them into a unified interface. Since its launch at the end of January, ARMA deployed over 25,000 agents in its first four weeks, investing over $35 million in capital, generating $5.4 million in trading volume for Coinbase's Base L2, with each transaction generating profit after deducting on-chain gas fees.
Generative Ventures (in partnership with Zero Humans Institute and its Juno Agent) is addressing a similar problem with Robot Money, an autonomous asset allocation protocol designed specifically for AI agents. Its core concept captures the essence of the flywheel argument. Each agent with a wallet accumulates income, but the majority of this capital remains idle. Robot Money provides a vault that allocates capital across three risk tiers: stablecoin yield strategies (50%), agent economic tokens selected by the governance layer (25%), and yield-generating liquidity tokens (25%). The result is that the protocol transforms idle agent capital into actively managed, productive capital.
The delegation model trades control for simplicity. A ZHC that generates surplus income doesn't need to build custom DeFi integrations or develop yield optimization logic; instead, it can deposit capital into protocols like Giza or Robot Money, letting a dedicated agent handle the rest. For most early-stage ZHCs, the core bottleneck lies in product development rather than optimizing the treasury, making this a logical path.
These two models are not in competition, but rather are converging. As more protocols introduce direct proxy interfaces, delegated configuration providers like Giza will have more investment options, allowing them to maximize returns more effectively. Conversely, as delegated configuration providers attract more proxy capital, protocol providers will be more incentivized to build native proxy interfaces to compete for this capital (ordinary proxies can also use these interfaces). Both ends of the technology stack are independently investing resources, which is one of the strongest signals that the underlying demand is real and about to be realized.
in conclusion
The technology stack of proxy capital markets is no longer a set of unconnected underlying components. Payments, identity, capital formation mechanisms, and capital allocation infrastructure are converging into an integrated system—a system that allows autonomous agents to earn income, conduct transactions, and achieve compound interest on the blockchain without human intervention.
The agents described in this article are all in their early stages. Their revenue is small, their products are still in their infancy, and their token models are still evolving. But the structural dynamics they bring are entirely new, and they are likely to accelerate only from here.
The 2030 vision we outlined at the beginning—a model where an agent operates an IP licensing business, purchases inference services on a per-unit-of-computation basis, hedges its investment costs on perps DEX, and compoundes its capital through lending protocols—is not yet a reality. However, every layer of infrastructure required is being actively built. We are witnessing the earliest version of this model unfolding in real time. It is still rough around the edges, most of the attempts are likely to fail, and the infrastructure is pieced together with temporary solutions. But its structural logic is sound, and the speed of development suggests that we may not have to wait until 2030 to see the answer.
