Circle founder Jeremy Allaire has outlined a future in which artificial intelligence agents and blockchain-based financial systems operate as one connected digital economy, with software programs capable of holding money, proving identity, signing contracts and completing commercial tasks without constant human input.
In an 89-page paper published on July 13 and titled The agentic economy, Allaire argued that AI and blockchain should not be viewed as separate technology trends. Instead, he described them as two parts of a single economic shift: AI lowers the cost of thinking and work, while blockchain lowers the cost of transferring value and enforcing digital records.
The paper presents a broad framework for how companies, money, labor and ownership could change if autonomous software agents become common participants in commerce. These agents would be able to perform business functions, coordinate with other software systems, use verified identities, and settle payments instantly across public blockchain networks.
Allaire’s argument is especially significant because Circle, the company he co-founded, is one of the largest issuers of dollar-backed digital currency through USDC. His paper places fully reserved, redeemable digital money at the center of a future machine-driven economy. While the paper reads partly as a technology thesis, it also raises wider questions about regulation, corporate structure, financial risk, labor income and economic power.
The paper does not claim that this system already exists in full. Rather, it describes a possible direction for the digital economy as AI tools become more capable and blockchain infrastructure becomes easier for software to use.
A proposed merger of AI and blockchain
Allaire wrote that AI is driving the “cost of cognition” toward zero. In simpler terms, software is becoming better at tasks that once required human judgment, research, writing, coding, planning or analysis. At the same time, he said blockchain networks are reducing the cost of economic coordination by allowing value and records to move across open, programmable systems.
In his view, the combination could create a new kind of economy where many commercial actions are handled by agents. These agents may not simply answer questions or automate basic office work. They could buy services, sell outputs, negotiate terms, borrow small amounts of working capital, and complete tasks on behalf of companies or individuals.
The paper frames this as a shift from human-centered internet activity to software-centered economic activity. Instead of people manually clicking through applications, signing documents, reconciling payments or managing routine workflows, agents could carry out those steps continuously in the background.
That would require more than advanced AI models. Allaire said agents need access to trustworthy money, reliable identity, audit trails and enforceable permissions. He argued that public blockchains can provide that foundation because they allow transactions and records to be verified by software without relying on a single private database.
How companies could change
One of the paper’s central claims is that companies may become less like fixed groups of employees and more like networks of specialized agents.
Allaire described future businesses as collections of modular digital workers. Some agents may handle sales, others may conduct research, manage customer support, write code, monitor compliance, negotiate supplier contracts or execute payments. A coordinating layer would assign goals, divide tasks, review outputs and combine results.
In this model, a company’s productive capacity would not depend only on its number of employees. It would also depend on the quality of its agents, the data they can access, the permissions they hold and the financial systems they can use.
The paper suggests that orchestration will become a major function. Companies will need systems that decide which agent should do which task, when a human must approve an action, how funds are released and how mistakes are handled. That could make management more technical, with business operations increasingly expressed through software rules and smart contracts.
Allaire did not suggest that human responsibility disappears. In fact, he argued the opposite. He wrote that autonomous systems need to be tied to identifiable people, companies or legal entities. A software agent may act on its own within defined limits, but someone must remain accountable for its actions.
Identity and accountability on public networks
The paper gives major importance to digital identity. For agents to participate in financial and commercial activity, they must be recognized as legitimate actors. That does not necessarily mean every agent would reveal all details publicly. But it does mean there must be a reliable way to connect an agent to an authorized person, company or institution.
Allaire argued that public blockchains can support this by recording credentials, permissions and transaction histories. If an agent signs a contract, spends funds or completes a task, the record can be checked later. This could make audits faster and reduce disputes over what happened.
However, the paper also notes a key limitation. Blockchain records can show that an action occurred, but they cannot automatically prove that the action was ethical, legal or properly authorized in every real-world context. A transaction may be transparent and still be harmful. A contract may be digitally signed and still be disputed. For that reason, human oversight, legal systems and governance remain necessary.
This is an important distinction. The paper does not present blockchain as a replacement for law. It presents blockchain as a programmable record and settlement layer that could make parts of commerce easier to verify.
Why digital money is central to the framework
Allaire’s framework depends heavily on digital currency that can move instantly and reliably between machines. He argued that the kind of money used by agents must be fully reserved, redeemable at face value and low risk.
That description aligns with the model used by regulated stablecoins, which are designed to maintain a stable value against a national currency such as the U.S. dollar. In this system, an agent could receive payment, hold funds, pay another agent, or settle a contract without waiting for traditional banking hours or manual processing.
Allaire wrote that instant settlement could play a role once filled by leverage in parts of the banking system. In traditional finance, fractional banking and credit expansion have helped money move through the economy, but they also introduce risk. In a machine-driven economy, he argued, speed itself could become a substitute for some forms of leverage. If money moves instantly and continuously, the same unit of value can support more activity over time without requiring unsecured expansion.
The paper also distinguishes base money from risk products built above it. Allaire argued that digital money used by agents should be as safe and simple as possible. More complex protections, such as escrow, insurance or conditional release of funds, could be layered on top when needed.
That approach reflects a broader theme in the paper: the base layer should be reliable, while more specialized services should operate above it.
Lending to software agents
The paper also explores how credit may work in an agent-based economy. Allaire suggested that underwriting could become cheaper because blockchain networks provide real-time, transparent data about activity, collateral, cash flows and repayment history.
This could allow small loans to be made to agents for specific tasks. For example, an agent may need working capital to buy computing power, acquire data, pay for software access or complete a customer order before receiving payment.
Allaire argued that such loans could be less risky in some cases because repayment may depend on predictable, machine-executed work rather than human behavior. If a task is clearly defined and payment terms are automated, a lender may have more confidence that the loan can be repaid.
Still, this part of the paper raises several unresolved questions. Software can fail, markets can shift, data can be wrong and tasks can produce poor results. Even if an agent behaves predictably, the surrounding economy may not. Credit risk would not disappear. It would change form.
Three layers of the agentic economy
Allaire described the proposed new economy as having three broad layers.
The first is base digital currency. This is the money agents use to transact. In his framework, it must be stable, fully backed, redeemable and available around the clock.
The second is programmable financial infrastructure. This includes payment rails, smart contracts, lending protocols, custody tools, identity systems and compliance mechanisms. These systems allow money and permissions to interact through code.
The third is agent execution. This is where AI agents perform work, buy services, sell outputs, manage tasks and coordinate with other agents.
Allaire argued that all three layers can operate globally as software. That creates tension with existing systems that are based on national borders, local labor rules and country-specific banking structures. If an agent in one jurisdiction buys services from an agent in another, using digital dollars on a public blockchain, enforcement becomes more complicated.
The paper suggests regulation may need to focus less on physical location and more on accountable entities. In other words, regulators may supervise the companies, individuals and institutions that create, deploy or control agents, even if the agents operate across borders.
A shift in the software market
The paper also predicts changes in how software is bought and sold. Today, many digital services are sold through subscriptions. Businesses pay monthly or annual fees for access to software, whether or not every feature is used.
In an agent-driven economy, Allaire said customers may increasingly pay for outcomes rather than access. An agent may pay a small amount to complete a single research query, generate one report, process one invoice, test one code function or verify one identity record.
That could push the software market toward pay-per-task pricing. AI models may become cost components inside larger business systems, while agents become the customer-facing layer that selects tools, manages workflows and pays for services.
This would also make micro-payments more important. Many existing payment systems are not well suited for tiny transactions, especially when fees are high or settlement is slow. Blockchain-based payments could make very small, frequent machine-to-machine payments more practical if transaction costs remain low.
Onchain companies and legal wrappers
Allaire used the term “onchain companies” to describe organizations that run much of their activity through programmable blockchain-based systems. These companies could hold funds, sign contracts, track records, distribute revenue and manage operations continuously.
He distinguished these from earlier token-based communities that often tried to operate without clear legal structures. In his view, many future onchain companies will likely keep a thin legal wrapper. That legal wrapper would connect the digital organization to courts, regulators, tax authorities and real-world accountability.
This is a practical point. A business may automate many functions, but it still needs to own assets, enter contracts, hire humans, comply with rules and resolve disputes. The paper suggests that future companies may become hybrid structures: partly legal entities, partly software networks.
Blockchain records could improve auditability by showing how money moved and which agents performed which actions. But the paper is careful not to claim that transparent records alone guarantee good governance. Companies would still need controls, approval processes and responsible managers.
Risks around labor and ownership
The paper’s most politically sensitive section deals with income and ownership. If AI agents handle more work, labor may account for a smaller share of economic value. That could reduce income for workers if new roles do not emerge quickly enough or if retraining fails to keep pace.
Allaire argued that falling labor income does not necessarily mean society becomes poorer overall. If productivity rises sharply, total output could increase. But the distribution of that output becomes the central issue.
If the gains from automation flow mainly to a small group of asset owners, inequality could grow. If ownership is spread more broadly, the benefits could be shared more widely. In Allaire’s view, ownership may become more important as a source of social standing and financial security if traditional work becomes less central.
The paper identifies several possible centers of power in the agentic economy. These include identity systems, authorization keys, digital currency issuers and the operators of critical infrastructure used by agents. Whoever controls these chokepoints could influence which agents can participate, which transactions are allowed and who captures the most value.
Governance remains an open question
Allaire concluded that the design of the agentic economy is not purely a technical matter. It is also political and economic. If agents and blockchains reshape production, payments and ownership, then rules around access, accountability and value distribution will matter as much as the software itself.
The paper suggests several possible mechanisms to broaden ownership and decision-making authority. These include participation-based ownership models and limits on token transferability designed to prevent control from consolidating among the largest holders.
Such ideas remain early and would likely face legal, technical and market challenges. But they show that Allaire’s thesis is not only about faster payments or smarter software. It is also about who controls the infrastructure of a more automated economy.
For traders and companies watching the sector, the paper is likely to intensify attention on stable digital payments, machine-to-machine transactions, AI identity tools and blockchain-based business systems. But it does not remove the major uncertainties around regulation, security, labor disruption and concentration of power.
The central message is clear: Allaire expects AI agents and programmable money to converge. Whether that convergence produces a more open and efficient economy, or a more concentrated and difficult-to-govern one, will depend on how the systems are built, regulated and owned.
Curious how autonomous agents already trade crypto? Explore AI copy trading as a real-world step toward Allaire’s agentic economy vision.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
