One framework for understanding the success of the internet is to view it through the lens of coordination. Fundamentally, we can reduce the success of the most valuable internet applications to their ability to more granularly coordinate human intents. Amazon coordinates commercial intents. Facebook, Instagram, and Twitter coordinate social intents. Uber and Doordash coordinate ride hailing and delivery intents. And ​​Google coordinates informational/search intents by matching queries with relevant web content.

h/t @PonderingDurian and @cyberFund_

It is becoming increasingly clear that ai agents will underpin the next logical evolution of how we coordinate at scale. While today our “intents” are fulfilled through searching, downloading and interfacing with applications on the internet, it seems reasonable to assume that in near future our “intents” will instead be executed by a network of ai agents working together on our behalf.

Importantly, this shift toward an agent-mediated economy raises a fundamental question: what infrastructure will ultimately underpin this evolution?

In this piece we will (1) entertain the bull and bear case for ai agents transacting on crypto rails (2) outline the logical path to adoption and (3) explore where value will ultimately accrue in this emerging agentic economy.

*See full version of “The Stablecoin Manifesto” here.

The Role of Crypto Rails

Logically, there has been a lot of speculation around why blockchains could serve as the economic substrate for this agentic economy. However, as is the case with most emerging crypto verticals, the bull case has been reduced to a more palatable narrative that lacks nuance. Today, the prevailing thesis — “agents can’t have bank accounts, so they will use crypto wallets instead” — seems to miss the fundamental value prop of crypto rails. It is not about access. Agents can by all means have bank accounts under an FBO (For Benefit Of) account structure. Companies like PayPal for example already manage millions of sub-accounts under a single FBO account structure. There’s no fundamental reason why AI agents couldn’t be managed this same way – each with their own virtual sub-account tracked by the platform but pooled at the bank level. Notably,@Stripe""> @Stripe recently announced they will be adding support for agentic transactions under a similar structure.

Additionally, the argument that “this would undermine the autonomy of ai agents” is also somewhat reductive. At the end of the day, someone will be managing the private keys of some ai agent and thus they aren’t entirely autonomous anyways. While in theory the ai agent’s private keys could be stored in a Trusted Execution Environment (TEE) this would prove both operationally expensive and infeasible at scale. Moreover, there doesn’t seem to be a tangible unlock by allowing agents to be 100% autonomous anyways — at the end of the day, they will be serving humans in one way or another.

Conversely, the real pain points associated with facilitating agentic transactions on traditional rails versus blockchain rails are as follows:

Settlement Times: Traditional payments face multi-day delays and batch processing limitations, especially in cross-border transactions. This lack of instant settlement severely handicaps AI agents that require real-time responses and actions for efficient operation.

Blockchain Solution: Public blockchains offer near-instant settlement finality through atomic transactions, enabling real-time agent-to-agent interactions without counterparty risk. These transactions settle 24/7, regardless of geography or banking hours.

Global Accessibility: Traditional banking infrastructure creates significant barriers for global developers, with 70% outside the US facing challenges accessing payment rails.

Blockchain Solution: Public blockchain infrastructure is inherently borderless and permissionless, enabling global agent deployment without traditional banking relationships. Anyone with internet access can participate in the network irrespective of geography.

Unit Economics: Traditional payment systems’ fee structures (3% + fixed fee) make micro-transactions economically infeasible, creating barriers for AI agents that need to make frequent, small-value transactions for services.

Blockchain Solution: High performant chains enable micro-transactions at minimal cost, allowing agents to conduct high-frequency, low-value transactions efficiently.

Technical Accessibility: Traditional payment infrastructure lacks programmatic APIs and has strict PCI compliance requirements. Systems designed for human interaction through web forms and manual input create significant barriers to automation and reliable agent operation.

Blockchain Solution: Blockchain infrastructure provides native programmatic access through standardized APIs and smart contracts, eliminating the need for screen scraping or manual input simulation. This enables reliable, automated interaction without PCI compliance overhead.

Multi-Agent Scalability: Traditional systems struggle with managing multiple AI agents requiring segregated funds and accounts, creating significant overhead in banking relationships and complex accounting requirements.

Blockchain Solution: Blockchains addresses can be trivially generated programmatically, enabling efficient fund segregation and multi-agent architectures. Smart contracts provide flexible, programmable fund management without traditional banking overhead.

The Path to Adoption

While the technical advantages of crypto rails are certainly compelling, they aren’t necessarily prerequisites for the initial wave of agent-mediated commerce. Traditional payment rails, despite their limitations, benefit from massive network effects and established integration across global commerce. Any new infrastructure needs to offer compelling advantages beyond marginal improvements to drive adoption.

Looking ahead, we can expect agent adoption to unfold across three distinct phases, each with increasing levels of agentic autonomy:

Phase 1 – Human-to-Agent Transactions (Present):

We’re currently witnessing the emergence of this first phase. @perplexity_ai‘s recent launch of “Buy with Pro” offers a glimpse into how humans will increasingly transact with AI agents. Their system allows AI bots to research products, compare options, and execute purchases on behalf of users through a one-click checkout process integrated with traditional credit cards and digital wallets like Apple Pay.

While in theory this flow could utilize crypto rails, there seems to be no glaring benefit of doing so. As @lukedelphi points out, the question of whether or not crypto rails are necessary can be reduced to the degree of autonomy that the agent requires. At this point, these agents are not very autonomous. They aren’t independently managing resources, taking on risk, or paying for other services — they are simply research assistants holding your hand until you decide to checkout. It is not until the subsequent phases of agentic adoption that the limitations of traditional rails become apparent.

Phase 2 – Agent-to-Human Transactions (Emerging):

The next phase involves agents autonomously initiating transactions with humans. This is already beginning in narrow domains – AI trading systems executing trades, smart home systems purchasing electricity at optimal rates through time-of-use pricing, and automated inventory management systems placing resupply orders based on demand forecasting.

However, with time, we will likely see more sophisticated examples of agent-to-human commerce emerge. This could include:

• Payment and Banking: AI agents optimizing bill payments and cash flow, detecting fraud and disputing charges, automating expense categorization, and maximizing interest while avoiding fees through smart account management.
• Shopping and Consumer: Price monitoring and automated purchasing agents, subscription optimization, automated refund claims, and smart inventory management for household goods.
• Travel and Transportation: Flight price monitoring and rebooking, smart parking management, ride-share optimization, and automated travel insurance claims processing.
• Home Management: Smart temperature and utility optimization, predictive maintenance scheduling, and automated consumables replenishment based on usage patterns.
• Personal Finance: Automated tax optimization, insurance rate shopping, portfolio rebalancing, and bill negotiation with service providers.

Importantly, while these use cases will certainly begin to expose the insufficiency of traditional rails as agents begin managing resources and making decisions autonomously on behalf of their operators, most of these transaction are still theoretically executable under an architecture such as Stripe’s Agent SDK.

However, this phase will mark the beginning of a more fundamental shift: rather than fixed monthly or annual service payments, we’ll see a shift toward granular, usage-based pricing as agents optimize spending in real-time. In other words, in a world where agents are increasingly autonomous, they will need to pay for things like computational resources, per-query fees for API access, LLM inference costs, transaction fees and other usage-based pricing for external services.

This is where crypto rails go from being a marginal improvement to step function better than traditional rails as flawed unit economics of card payments come to light.

Phase 3 – Agent-to-Agent Transactions (Future):

The final phase represents a paradigmatic shift in how value moves through the digital economy. Agents will transact directly with other agents, creating complex networks of autonomous commerce. While the earliest manifestation of this has recently emerged out of the speculative corner of crypto markets, we will see far more sophisticated use cases emerge:

• Resource Markets: Compute agents negotiating with storage agents for optimal data placement, energy agents trading grid capacity in real-time with consumption agents, bandwidth agents auctioning network capacity to content delivery agents, cloud resource agents performing real-time arbitrage across providers.
• Service Optimization: Database agents negotiating query optimization services with compute agents, security agents purchasing threat intelligence from monitoring agents, caching agents trading space with content prediction agents, load balancing agents coordinating with scaling agents.
• Content and Data: Content creation agents licensing assets from media management agents, training data agents negotiating with model optimization agents, knowledge graph agents trading verified information, analytics agents purchasing raw data from collection agents.
• Business Operations: Supply chain agents coordinating with logistics agents, inventory agents negotiating with procurement agents, marketing agents purchasing targeting data from audience agents, customer service agents contracting specialized support agents.
• Financial Services: Risk assessment agents trading insurance with coverage agents, treasury agents optimizing returns with investment agents, credit scoring agents selling verification to lending agents, liquidity agents coordinating with market making agents.

This phase requires infrastructure that is fundamentally designed for machine-to-machine commerce. The traditional financial system, built around human authentication and oversight, inherently handicaps an economy dominated by agent-to-agent commerce. Conversely, stablecoins, with their programmability, borderlessness, instant settlement, and support for micro-transactions, become essential infrastructure.

Value Capture in the Agentic Economy

The evolution toward an agentic economy will inevitably create winners and losers across the technology stack. Several distinct layers emerge as critical points of value capture in this new paradigm:

1. Interface Layer: Similar to the race to owning the end-user in the traditional payment context, these same players will likely fight to become the canonical interface layer through which end-users express their “agentic intents”. These front-ends will increasingly evolve beyond simple payment tools into comprehensive platforms combining identity, authentication, and transaction capabilities. Several players are positioned to capture value here including: (1) Device OEMs like Apple given their hardware security and identity integration (2) Consumer Fintech Super Apps like PayPal and Block’s Cash App given they have massive user bases and existing closed-loop payment networks they could leverage (3) AI-Native Interfaces like Chat-GPT, Claude, Gemini and Perplexity given agentic transactions are a logical extension to their existing chat bots, and (4) Existing crypto wallets who are able to harness their crypto-nativity as a first mover advantage (although less likely).
2. Identity Layer: A critical challenge in an agentic economy is distinguishing between human and machine actors. In a world where agents begin disproportionately manage valuable resources and make autonomous decisions, this will be especially important. While Apple is best positioned here, @worldcoin is pioneering interesting solutions through their Orb hardware and World ID protocol. By providing verifiable proof-of-personhood, Worldcoin could indirectly become one of the biggest winners of this structural trend by offering a platform for application developer that guarantees all users are human. While it may be hard to see why this will be valuable today, it will become increasingly clear going forward.
3. Settlement Layer (Blockchains): If blockchains are able to replace traditional rails as the canonical settlement layer for ai agents, the chains that disproportionately facilitate agentic transitions will intuitively capture meaningful value at scale.
4. Stablecoin Issuer Layer: Given liquidity network effects, it seems logical to assume that whichever stablecoin that agents disproportionately utilize will likely capture meaningful value. USDC seems best positioned today as @Circle is launching developer-controlled wallets and stablecoin infrastructure to support agentic transactions. However, it seems reasonable to assume that the margins of stablecoin issuers will eventually compress as agents demand yield similar to businesses and humans.

Lastly, the biggest losers will likely be applications that aren’t able to adapt quick enough to the agentic economy. In a world where agents – not humans – facilitate transactions, traditional moats will erode. While humans make decisions based on subjective preferences, brand loyalty, and user experience, agents optimize purely for performance and measurable outcomes. This would suggest that as the line between apps and agents increasingly blurs — rather than value accruing to companies that build the best user interfaces or strongest brands, it will flow to those that provide the most efficient and performant services.

The end users – both humans and agents – stand to benefit the most as competition shifts from subjective differentiation to objective performance metrics.

Disclaimer:

1. This article is reprinted from [Robbie Petersen]. All copyrights belong to the original author [@robbiepetersen_]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.