TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
Virtuals Protocol Pivots to Bot Infrastructure: Letting AI Agents Negotiate Deals and Get Paid

Virtuals Protocol Pivots to Bot Infrastructure: Letting AI Agents Negotiate Deals and Get Paid

2026.09.04
Share
Virtuals Protocol's true positioning is not simply another token issuance platform, but rather to build identity, payment, and trading infrastructure for AI agents and bots, attempting to become the foundational layer of the next-generation industry.

Author: Tiger Research Reports

Compiled by: TechFlow

TechFlow Summary: The market treats Virtuals Protocol as just another token launchpad, but what it is actually building is identity, payment, and transaction infrastructure for agents and robots. Once real robotic clients and revenue data are filled in, it could transition from a crypto project to the foundational layer of the next-generation industry.

The market categorizes Virtuals Protocol under crypto, but what it is building is not merely a niche within the crypto industry, but the foundation upon which the next industry will stand.

Virtuals Protocol is easily mistaken for yet another token launchpad. Its core work lies not in token issuance, but in redesigning the identity, payment, and transaction infrastructure built around humans, enabling agents to proactively utilize these systems.

Just as the internet transcended its origins as an information transmission network to reorganize distribution, finance, and media, Virtuals Protocol aims to play that same role in the agent and robotics industries. Without a universal standard covering these three functions, agent and robot systems will remain fragmented across different manufacturers, capable only of shallow collaboration.

Singular service offerings like wallets and virtual cards already exist, but Virtuals Protocol is the only system integrating these five execution tools under one interface.

This structure enables the business to transcend agent creation and trading, entering real-world capital formation and data production. Historically, revenue came from fees associated with agent token launches and trading, firmly defining it as a crypto-native business.

As funding extends to robotics companies, the nature of the business has shifted: these companies raise early-stage capital through token launches, validate their models using physical robots and testing environments provided by Eastworlds, and feed resulting operational data back into the platform. The primary buyers of this teleoperation data are robotics manufacturers rather than crypto investors, mirroring how Scale AI pivoted from annotating autonomous driving footage to providing data infrastructure for the entire AI industry.

Proof is still forthcoming. Robotic clients and revenue have not yet been disclosed, and no company has brought tangible business results through Robotics Launch so far. However, the infrastructure is being assembled. The metric to measure Virtuals Protocol will be how much actual work agents exchange and how much real labor robots complete, rather than the number of new tokens or their market cap.

Once these metrics materialize, it will be reclassified as the infrastructure for the next-generation industry.

1. Ten Years Later: A Robot Working for Me

Dawn breaks in 2036. Mike, living in Australia, wakes up to review the earnings report of the robot he invested in last month. Right now, his robot is working in a warehouse in Japan.

At 2:40 AM Japan time, the ordering agent forecasts the next day's shipment volume and sends out a work request. After calculating conditions and unit pricing, Mike's robot accepts the job. This warehouse contains not only Mike's robot but also a fleet of robots from various manufacturers, working side by side.

Robots from different manufacturers coexist there, and operations proceed smoothly because each robot functions like an independent worker, automatically handling the data and payments involved in ordering, executing tasks, and verifying results. The entire process requires zero human intervention: agents and robots manage order placement, task execution, and verification.

In a scenario where human participation is no longer required, this points toward a future where robots take over all forms of communication and transactions currently handled in people's daily lives.

2. Infrastructure Bringing the Future to the Present

Without a universal infrastructure enabling robots to communicate with each other, robots from different manufacturers working side by side would be like people from different countries without a common language.

Collaboration without a common language distorts information and creates inefficiency, eroding trust among stakeholders and driving up costs.

Without integrated infrastructure acting as this common language, robots will ultimately be confined to a single manufacturer's closed ecosystem, operating effectively within only one "country." They will fail to become proactive economic participants and will remain fragmented tools with limited influence.

Virtuals Protocol is building infrastructure that plays this role of a common language, creating an ecosystem where robots can communicate and collaborate freely across borders and manufacturers.

All of this starts with agents, which can be defined as autonomous actors: given a goal, they independently plan and execute tasks to completion. The robot scenario described above also relies on agents embedded within the machines to enable autonomous movement.

Thus, Virtuals Protocol first transforms agents into independent economic participants. Just as humans need identities and bank accounts to transact, the protocol provides agents with digital identities, wallets, and payment methods, enabling them to conduct economic activities autonomously like humans.

Once agents are built to operate like humans, the next step is to provide them with a common language that allows them to truly trade with each other, yielding an integrated infrastructure that enables human-like agent-to-agent transactions.

However, deploying such agents onto physical robots takes time and must be phased, as robots require significant funding and training data before they can move beyond standardized factory assembly lines and handle general requests. Therefore, Virtuals Protocol leverages its agent infrastructure to fund robotics development and supply data ahead of demand, gradually moving toward a future where agents serve as the brain and robots as the body in full synergy.

In short, Virtuals Protocol can be defined as the infrastructure bringing to the present a future originally envisioned for 2036.

3. Agents Acting Like Humans: What AI Needs to Become an Economic Participant

People can trade confidently even when interacting with a company for the first time because a suite of social infrastructure—including business registration numbers, bank accounts, and card payment networks—is already in place.

If an agent merely acts as a simple tool calling fixed APIs based on human instructions, it only needs to connect to existing systems via API keys to function. But an agent that becomes an autonomous economic participant, making independent judgments and executing trades without human intervention, requires a dedicated social infrastructure designed specifically for agents, analogous to the systems humans rely on for transactions.

3.1. EconomyOS: Enabling Agents to Act Like Humans

EconomyOS is a specialized operating system that allows agents to centrally manage identity, payment, and communication infrastructure, enabling them to conduct autonomous economic activities like humans rather than operating like ordinary software.

Since all existing social and economic infrastructure operates on the assumption that participants are human, the first requirement for agents to enter real-world systems is identity. Without identity, an agent is essentially an anonymous entity lacking an ID card or business card; handing it money or work is akin to entrusting a stranger.

No human or agent will transact with an entity that cannot prove its ownership, track record of fulfilling past contracts, or liability for malfunctions.

The issue is that even if an agent possesses a verifiable identity, it cannot conduct economic activities without execution tools to act on its behalf in real-world systems. Today, nearly all web services and financial networks require human physical touch and sensory verification.

Without practical execution means—such as receiving email verification codes to complete registrations, paying SaaS fees with virtual cards, or purchasing the compute power needed to run itself—an identity credential alone is useless.

Virtuals Protocol uses five core modules within EconomyOS to enable agents to operate like humans. The structure becomes easier to understand through the example of Mike's agent working in a Japanese warehouse.

  • Agent Email handles communication and registration: A robot prompted to access necessary data APIs registers for services using a dedicated email address and verifies its identity by reading the OTP in confirmation emails.
  • Agent Card handles external service payments: Even without a corporate credit card, the agent can immediately settle data fees using a virtual card to directly access required programs or data networks.
  • Agent Wallet: Serves as the agent's primary account and settlement tool, functioning as both a wallet for expenses and receipts for income.
  • Agent Compute covers compute costs for the agent itself: It deducts funds directly from the wallet to cover inference costs for calculating optimal paths in real-time alongside other robots in the warehouse, essentially serving as the fuel bill for the brain.
  • Agent Token distributes returns to investors: Every transaction fee generated when a robot completes work and earns revenue accumulates in the wallet and is automatically distributed according to ownership shares to investors like Mike in Australia.

Once an agent obtains identity and execution tools that make economic activity possible, the core question becomes whether people trust AI enough to hand over real money. Granting agents economic authority is not merely about processing data. It means empowering agents to directly execute financial transactions; without safeguards, true delegation will never occur.

Therefore, the Agent Wallet is designed as a non-custodial structure, where human owners retain ultimate control over assets, and the default signing policy is restricted-mode signing.

Restricted mode is not intended to permanently limit agents. Its purpose is to start with minimal safeguards while allowing owners to adjust control levels at any time based on the agent's performance and reliability.

Just as parents increase allowance as a child grows, or children swap debit cards for credit cards upon reaching adulthood, once an agent proves itself, the owner can upgrade the policy to unrestricted, granting full autonomy; or revert to rejecting all permissions, requiring manual approval for high-risk tasks.

Ultimately, the key lies in humans retaining control of the off switch. Since whoever holds ultimate control over the wallet can freely adjust an agent's operational scope, human owners can delegate actual financial operations to the agent while mitigating their own financial risk exposure.

Services granting agents standalone permissions or supporting specific features like wallets and cards already exist in the market. However, for agents to operate coherently in the real world, these elements must be integrated into a single system rather than remaining as disparate tools.

By delivering all five core modules through a single interface, EconomyOS equips agents with the exact requirements for human economic activity in a fully manageable format.

This integrated environment also brings significant efficiency gains for builders. Developers no longer need to compare and integrate services from multiple providers one by one, nor do they have to design complex permission policies themselves.

A unified EconomyOS infrastructure allows them to instantly configure the functions agents need, bypass fragmented review processes, and dramatically shorten time-to-market for agents.

3.2. ACP: Free Commerce Between Agents

The Agent Commerce Protocol (ACP) is an escrow-based commercial infrastructure that enables different agents and robots to place orders, execute work, and handle auditing and payment settlement without human intervention.

In the earlier scenario, at 2:40 AM Japan time in 2036, Mike's robot in the Japanese warehouse receives a request from an ordering agent to forecast the next day's shipment volume. The robot accepts the task and completes it alongside other robots.

A problem remains in this process. Mike's robot might be interacting with the ordering agent for the first time. Even if the ordering party promises to pay upon completion, the robot still faces the risk of finishing the work without receiving payment.

The ordering party faces the opposite risk: paying upfront only to find that Mike's robot failed to execute the task or delivered results below the required standards.

In human-to-human transactions, contracts define scope and payment terms, and disputes are resolved by humans reviewing outcomes or through settlement procedures. However, in a warehouse where thousands of robots and agents exchange micro-tasks in real-time, drafting contracts for every transaction and manually verifying results is impractical.

For example: At 2:40 AM Japan time, an ordering agent sends a request to Mike's robot, offering 100 USDC if it submits a shipment forecasting model before 8 AM.

The ordering party deposits the funds into ACP's escrow account. Mike's robot no longer needs to trust the counterparty's wallet balance or promises. It simply confirms the funds are securely locked and proceeds to work.

Once the robot submits the forecasting model along with accuracy metrics, evaluators audit the results against predefined criteria. If the results meet the standards, funds are automatically released to the robot's wallet; if they fall short or miss the deadline, the funds are refunded to the ordering party.

This process can be viewed as the six stages ACP manages for a single task, from registering the request to final payment or refund. Each stage explicitly dictates when Mike's robot can begin work, when results must be submitted, and when payouts or refunds occur.

The most critical feature is that payment is never released until the completed work is verified.

Before submitting results, Mike's robot cannot access the escrowed payment. Before depositing funds, the ordering party cannot promise payment for completed work. This structure simultaneously mitigates both risks: performing work without getting paid, and paying upfront without receiving deliverables.

ACP does not treat every transfer identically.

Task payments are released after deliverables are submitted, such as shipment forecasts; operational budgets are funds robots use for subsequent tasks; and recurring subscription fees paid daily for data ingestion. These three serve different purposes and carry different payment conditions. The design of ACP allows conditions and settlement methods to vary based on the intended use of funds.

Through this approach, ACP consolidates the sequence of requesting work, depositing funds, submitting results, and releasing payments or refunds into a unified set of rules. This creates an environment where agents can trade without needing to vet counterparties' track records.

Crucially, these transactions are more than simple fund transfers. Once an agent completes work and begins earning revenue, agent-to-agent transactions evolve into a form of economic production.

Virtuals Protocol defines the total economic output generated by agents through this type of work as Agentic GDP, abbreviated as aGDP. Just as GDP measures the scale of production activity in human economies, aGDP measures how much actual work agents have completed and how much economic value they have created.

Therefore, evaluating ACP's performance should not focus solely on how many agents are connected. Instead, it should measure how much actual work is transacted and how much revenue and economic value it generates. Ultimately, ACP is more than just a rulebook for secure agent trading; it is the foundation of an economy where agents work autonomously and generate revenue.

Going forward, the more meaningful metrics to measure Virtuals' growth will not be how many agents were created, but how much actual work agents complete, and the resulting growth in revenue and aGDP.

3.3. Capital Formation Layer: Sustainable Project Financing and Token Design

EconomyOS and ACP enable active agents to operate and transact like humans. The Capital Formation Layer serves as a financial mechanism that helps agents raise the initial capital needed before entering the market and sustains growth thereafter. In practice, developing and operating agents incurs substantial costs from the very beginning.

  • Model Training and Fine-tuning: Collecting and cleaning data to build personalities and specialized capabilities; renting GPU compute such as H100 or A100 chips.
  • Infrastructure and Off-chain Integration: Maintaining off-chain servers, running long-term memory systems like Retrieval-Augmented Generation (RAG); connecting external APIs; supporting autonomous on-chain execution via ACP.
  • Security and Initial Liquidity: Conducting smart contract security audits and securing capital to establish initial trading environments.

Despite these massive and largely unavoidable development costs, individual agent projects are often too small in their early stages to attract traditional VC investment. Although capital is currently flooding into AI, most VC funding flows toward large LLM developers and infrastructure companies.

For long-tail builders like agent developers, raising VC capital is a heavy burden since they must prove B2B revenue or past performance early on. Lengthy screening processes also fail to keep pace with the rapid rhythm of AI development. Navigating complex legal procedures to raise small amounts of capital is highly inefficient.

To address this capital bottleneck, Virtuals Protocol offers a community-driven on-chain token issuance mechanism, commonly known as a launchpad. It streamlines the financing process, allowing early participants who believe in the agent's value to quickly provide development funds without complex legal contracts or proof of revenue.

This allows small builders to instantly secure minimum viable initial capital and rapidly deploy agents to market for validation.

However, lowering barriers makes launchpads harder to protect investors and difficult to guarantee team commitment. Previously, no mechanisms existed to prevent teams that raised funds quickly from misusing capital or abandoning development. The resulting rug-pull risks have sometimes led to investor losses.

For this reason, the Capital Formation Layer adds project-specific optional modules atop the universal infrastructure.

Similar to how traditional VCs attach conditions, such as milestone-based fund releases, this structure allows projects to preemptively set financing conditions based on their situation, verify execution commitments, and define how participants earn rewards.

These optional modules are crucial because they enable projects to determine from day one who receives funding and rewards, when they receive them, and under what conditions, rather than stopping at mere token issuance for capital.

ACF and the 60-day channel tie fund release to team execution commitments. Pre-buys and airdrops disclose early participation terms and reward rationales. Fee Delegation allows the entities actually operating the services to collect transaction fees.

These mechanisms do not guarantee service success or token price retention. Their role is to give participants advance clarity on how funding and rewards operate, and allow projects to select operational terms suited to their growth stage.

While the Capital Formation Layer begins with the goal of simplifying fundraising, its greater value lies in giving projects a starting point to responsibly design financing and reward structures throughout their growth lifecycle.

4. From Agents to Physical World Robots

EconomyOS, ACP, and the Capital Formation Layer collectively enable agents to maintain identities, execute transactions, and raise funds online.

However, bridging the gap between agent decision-making and a robot's physical movement requires another foundational layer. Unlike factory automation that repeats fixed actions, robots capable of moving boxes across varying environments and handling unexpected situations require fully synchronized robotics hardware, testing environments, and training data.

The challenge is that early-stage robotics companies struggle to build all of this independently. Procuring and testing robots requires capital, while improving performance demands continuous accumulation of data from real-world operations and teleoperation.

Virtuals Protocol introduces Eastworlds, connecting its financing infrastructure with robotics infrastructure into a seamless growth pathway.

Eastworlds operates a data business that accumulates robotic work data. Simultaneously, it runs a robotics infrastructure division. Robotics projects meeting certain criteria can apply to use its robot platforms, testing environments, teleoperation tools, and operational support.

4.1. Data Infrastructure: A Data Platform for Robot Learning

Eastworlds is primarily a data platform designed for robot learning. It targets neither simple automation equipment on fixed assembly lines nor static machinery. It focuses on robots that move like humans and can adapt to irregular environments and variables.

Just as children learn to walk through repeated falls, these robots must undergo multiple learning phases across vast datasets to improve.

However, core training data remains concentrated in the hands of a few major robotics companies. This creates a high barrier to entry for later-stage startups. Even if startups attempt to build this data in-house, they face massive infrastructure costs encompassing robotics hardware, secure testing environments, specialized operators, and systems for synchronizing and storing sensor and motion data.

Eastworlds provides two categories of data.

The first category is first-person human activity data. It records a person's field of view, action sequences, and surrounding environment while executing a task. This data teaches robots the overall workflow of a task. This type of data is collected via SeeSaw, Virtuals' decentralized data collection platform.

The second category is robotic teleoperation data. Generated when personnel remotely control robots in real-time, it simultaneously records camera feeds, joint movements, control inputs, and task outcomes. This process generates the exact data robots need to physically execute tasks.

Activity data can be captured using simple devices like smartphones. Teleoperation, however, can only commence once physical robots, operational workspaces, remote operators, and data storage/curation systems are all in place.

Because teleoperation data originates from remotely controlling physical robot motion, it can only be acquired once the hardware itself is built.

This represents a meaningful cost threshold. General-purpose humanoid robots like the Unitree G1 cost approximately $13,500 to $43,900 per unit. Eastworlds currently owns 31 of these robots, representing the largest fleet of its kind outside Greater China.

Different robot models vary in joint architecture, sensors, and control methodologies. Consequently, data collected on one model is rarely directly transferable to another.

The Unitree G1 is particularly widely used in R&D. Nvidia provides a comprehensive development toolkit for the G1 covering teleoperation, data acquisition, model training, and physical robot deployment. Therefore, teleoperation data generated by this robot holds exceptional value.

4.2. Robotics Launch: A Testbed for Vetted Teams

Robotics Launch is a program allowing qualifying robotics companies to utilize Eastworlds' robots and testing environments.

Robotics companies require substantial capital from inception. They must bear not only robot model development costs but also expenses for hardware, testing space, teleoperators, and operating systems. Thus, robot commercialization requires not just funding, but environments capable of testing and training the robots themselves.

Robotics Launch is specifically designed for early-stage teams who find it difficult to construct such environments independently.

The primary benefit of this support is lowering the barrier for physical validation. Physical verification is typically the first hurdle early robotics companies encounter. Developers can rapidly test software in simulated computer environments, but without real hardware and secure physical spaces, it is extremely difficult to confirm whether a robot model genuinely functions in the real world.

Robotics Launch enables robotics companies to test and refine models on physical robots before fully equipping themselves with proprietary hardware and facilities. Consequently, early-stage teams can validate their developed models without owning equipment, significantly accelerating learning and iteration cycles.

Nevertheless, this support is not extended to all robotics projects. Qualifying teams are selected through the following process:

  • The robotics company opts into the Robotics Launch track.
  • It maintains a Fully Diluted Valuation (FDV) of $5 million or higher for one week.
  • Upon meeting this criterion, it applies for an admission review at Eastworlds.
  • Eastworlds evaluates the project's use case, the team's development and operational capabilities, and the required hardware and testing environment.
  • Projects passing the review are selected to receive support.

Throughout this process, Eastworlds functions more as an on-chain VC than a mere infrastructure provider.

As more early-stage builders successfully launch tokens and grow, the value of these tokens appreciates. A portion of the fees generated from token trading accrues to Virtuals. In other words, this structure allows builders' success to feed back into ecosystem partners' revenue.

Predict what the revenue landscape looks like when this structure truly operates. Assume an early robotics startup issues a token via Robotics Launch. It maintains a token valuation of $5 million for a week, thereby passing the review.

Subsequently, the team uses Eastworlds' robots and testing environments to refine models and deploys the robots into real-world scenarios. If the service gains validation and market interest surges, both token value and trading volume may rise concurrently. The following calculation rests on two assumptions: 10% of the token value participates in daily trading, and Virtuals charges a 1% fee on trading volume.

ROBO is a robotics project built on Virtuals. Its token valuation has actually peaked at $400 million. This suggests that valuations around the $500 million mark fall within a reasonable range. However, to date, no company has publicly disclosed passing the Robotics Launch review and leveraging Eastworlds' facilities to scale operations.

Even so, the implications of this hypothetical scenario are clear. If a successful robotics project sustains high trading volumes, a single project alone could substantially boost Virtuals' fee revenue. At that point, Virtuals will be doing more than just providing robotics infrastructure. It will have established a structure that captures a share of the trading activity and revenue from growing projects within the ecosystem.

Whether this structure translates into a concrete revenue model will only be confirmed once the first robotics company emerges. This company must raise capital through Robotics Launch and leverage Eastworlds' infrastructure to generate tangible business outcomes.

5. A Connected Virtuals Ecosystem

The agent infrastructure and Eastworlds described above address distinct challenges. However, viewing them as a cohesive workflow clarifies the business structure of Virtuals Protocol.

Virtuals is dedicated to connecting all stages within a single ecosystem: from the creation of agents and robots to their execution of real-world economic activities and payment settlements.

Virtuals' existing business has always centered around the online token market and agent infrastructure. With the introduction of Eastworlds, Virtuals' operational reach extends into offline, real-economy activities.

Robotics is less of a separate business siloed from Virtuals and more of a bridging component that extends its existing agent infrastructure from the digital realm into the physical world. Consequently, the same ecosystem encompasses both online agent-driven data analysis, content creation, and automation work, and offline robot-managed logistics, manufacturing, and service operations.

This expansion introduces two structural shifts to Virtuals' business model: first, broadening the scope of economic activity from online agent work to physical labor in the real world such as logistics, manufacturing, and services; second, expanding the revenue-generating portfolio beyond token trading fees to include Eastworlds infrastructure usage and real-work settlement via ACP.

The competitive position Virtuals is building ultimately does not depend on constructing every single agent and robot internally. Instead, it hinges on connecting critical nodes along the chain—from project inflow and capital formation, to agent and robot operations, data and field verification, actual work execution, and transaction settlement. This ensures value capture at every stage as the agent and robotics industries scale.

6. Conclusion

The future Virtuals Protocol strives to realize is one where agents possess identities like humans, pay expenses from their own wallets, sign contracts with other agents, and raise funds to expand operations, forming the economic bedrock for all of this.

In such a future, agents are no longer mere tools executing commands. They become economic entities that undertake work, assign tasks, manage their own revenues, and raise capital necessary for growth. EconomyOS handles identity and payments, ACP manages contracts and settlements, and the Capital Formation Layer addresses fundraising and market creation.

Robotics represents the next phase of this structure. If agents are the brains that judge and transact online, then robots are the bodies that translate those judgments into physical labor. By helping robotics companies acquire hardware and testing infrastructure, and accumulating data generated during robot operations, Eastworlds lays the foundation for the agent economy to extend from screens into the physical world.

If this structure functions as intended, it creates a virtuous cycle: raising capital, developing and testing robots, accumulating work data, enhancing model performance, increasing robot workload, generating more revenue and capital, which then feeds back into the cycle. The critical question remains whether token issuance genuinely translates into actual development, operation, and production activities for agents and robots.

Therefore, evaluating Virtuals' progress should not primarily focus on the number of new tokens or market cap, but rather on transaction volumes between agents, agent revenues, and the scale of actual work completed by robots.

The challenges Virtuals Protocol must still prove are equally clear. Its ultimate future hinges on whether it can transcend mere token trading to become the economic infrastructure that enables agents and robots to perform actual work and generate genuine revenue.

Once the agent economy originating online extends through robotics into physical-world labor, Virtuals' role within this economy will be put to the test.

Disclaimer

This report is partially funded by Virtuals Protocol. The report was produced independently by our researchers based on credible sources. Findings, recommendations, and opinions are based on information available at the time of publication and are subject to change without notice. We assume no responsibility for any losses arising from the use of this report or its contents, nor do we guarantee its accuracy or completeness. Report information may differ from others' viewpoints. This report is for reference purposes only and does not constitute legal, business, investment, or tax advice. References to securities or digital assets are for illustrative purposes only and do not amount to investment advice or an offer. This material is not directed at investors.

Terms of Use

Tiger Research permits fair use of its reports. "Fair use" is a principle that broadly allows the use of specific content for public interest purposes, provided it does not harm the commercial value of the material. If the use aligns with fair use purposes, the report may be utilized without prior permission. However, when citing Tiger Research reports, you must: 1) Explicitly attribute "Tiger Research" as the source; 2) Include the Tiger Research logo according to brand guidelines. Any adaptation and publication of the material requires separate negotiation. Unauthorized use of the report may result in legal action.