From OpenClaw to the $25 billion RWA market: How AI agents are quietly taking over on-chain assets

In March 2026, Illia Polosukhin, co-founder of the NEAR Protocol, made a seemingly simple yet profound statement in an interview: “Blockchain users will be AI agents.” He painted a future vision: AI will become the front-end interface for all online transactions, while blockchain recedes into the background as a trusted backend infrastructure. Humans will no longer need to directly operate wallets, browse block explorers, or verify transaction hashes; these complexities will be fully abstracted by AI agents.

Almost simultaneously, the open-source AI agent project OpenClaw released version v2026.3.7-beta.1, which natively supports GPT-5.4. This project, with over 280,000 stars on GitHub, has rolled out two major updates in two days. The official changelog included a self-deprecating yet confident remark: “We fix more issues than we create—that’s progress.” This update not only introduced pluggable context engines but also enhanced security mechanisms and deployment capabilities—OpenClaw is evolving from an experimental agent framework into a true “agent operating system.”

Meanwhile, another seemingly unrelated piece of news is circulating in the crypto community: data from RWA.xyz shows that the on-chain value of tokenized real-world assets (excluding stablecoins) has surpassed $25 billion, nearly quadrupling from about $6.4 billion a year earlier. The on-chain scale of six major asset classes—U.S. Treasuries, commodities, private credit, institutional alternative funds, corporate bonds, and non-U.S. government debt—has all exceeded the $1 billion threshold.

These events, occurring within the same month, are no coincidence. They collectively point toward an emerging paradigm shift: as AI agents begin to autonomously interact with blockchain, and as the on-chain asset scale supports a “proxy economy,” the operation mode of RWAs will shift from “manual management” to “AI autonomous management.” This is a significant industry transition that warrants serious attention.

1. AI is shifting from “co-pilot” to “main pilot”

To understand the depth of this shift, one must first recognize the fundamental change in AI’s role.

In recent years, AI has primarily played a “co-pilot” role in public perception—assisting humans with writing emails, planning trips, generating code—but always in a passive response mode. Users issue commands, AI executes them, and humans complete the task in a closed loop. In this model, AI is a tool, and humans are the main actors.

However, the latest release of OpenClaw provides a window into how this relationship is loosening. On March 7-8, OpenClaw released versions 2026.3.7 and 2026.3.8, with core updates focused on four areas: model capability upgrades, agent architecture evolution, deployment optimization, and security/reliability enhancements.

Of particular interest to developers is the pluggable Context Engine. This mechanism allows developers to freely attach RAG or lossless compression algorithms, solving the “forgetfulness” problem in long conversations and paving the way for long-term autonomous operation. Additionally, ACP binding support for restart recovery means that even after server reboots, agents can “remember” previous interactions and context, enabling truly persistent services.

Behind these technical details lies a significant trend: AI agents are gaining “persistence” and “autonomy.” They are no longer one-off dialogue products but digital entities capable of continuous operation, learning, and task execution.

Polosukhin’s prophecy highlights potential application scenarios: “AI will be at the front end, with blockchain serving as the backend. The goal is to hide the entire blockchain from your AI—our possession of block explorers is actually a failure because we haven’t abstracted this technology.”

He envisions future AI agents directly interacting with blockchain protocols, autonomously handling payments, managing assets, coordinating services, and even participating in governance votes. Humans will only need to converse with AI, telling it “optimize my asset allocation” or “vote on that proposal,” while the agent completes the work on-chain.

This is not science fiction. OpenAI and Paradigm’s collaboration on EVMbench is already testing AI agents’ ability to detect, patch, and exploit smart contract vulnerabilities. Circle and Stripe are racing to build stablecoin payment infrastructure for AI agents, with Stripe’s x402 USDC payment feature on Base supporting autonomous settlement among agents. Decentralized AI infrastructure protocols like 0G and Alverse’s “Web4.0 marketplace” enable AI agents to mint and trade digital assets using encrypted proxy IDs.

An on-chain economy composed of AI agents is moving from concept to reality.

2. From issuance to governance, every stage of RWA is being rewritten

When AI agents become “users” of blockchain, the entire lifecycle of RWAs—issuance, trading, management, and governance—will be systematically transformed. This is not just efficiency optimization but a fundamental paradigm overhaul.

Asset issuance: from “manual due diligence” to “real-time verification”

Traditional RWA issuance involves multiple manual interventions—lawyers, auditors, appraisers. For example, in real estate tokenization, project teams need third-party valuation reports, property rights investigations, cash flow audits, often taking months and incurring high costs.

AI agents can revolutionize this process. By integrating IoT devices, on-chain credit scores, third-party APIs, they can verify asset status in real time. For instance, once a batch of goods’ ownership certificates are on-chain, and insurance and customs documents are verified, AI agents can automatically trigger tokenization, generating corresponding RWA tokens for investors. This compresses the process from months to minutes, with minimal human intervention.

Trade execution: from “instruction response” to “strategy gaming”

Currently, RWA trading relies heavily on manual orders or simple smart contract triggers. Investors switch between platforms, compare prices, assess liquidity, and manually execute trades.

AI agents can execute complex strategies. They can monitor multiple markets for arbitrage opportunities, perform cross-chain arbitrage automatically; predict asset price movements based on macroeconomic data (interest rate decisions, inflation reports), and preemptively adjust holdings; or automatically execute stop-loss or hedging when preset risk thresholds are triggered. Moreover, multiple AI agents competing in the same market will generate complex dynamics—challenging for humans to simulate but potentially increasing market efficiency.

Asset management: from “monthly reconciliation” to “continuous monitoring”

Management during RWA’s lifecycle is often overlooked. Rent collection, interest payments, collateral monitoring, profit distribution—these daily operations depend on manual reconciliation and collection, which are inefficient and error-prone.

AI agents can provide 24/7 monitoring. They can automatically distribute cash flows to investors’ wallets; send alerts or initiate liquidation when collateral values fall below thresholds; and handle early redemptions or renewals based on preset rules. For investors, this means greatly improved transparency and timeliness in asset management.

Governance participation: from “low voting turnout” to “algorithmic democracy”

Tokenized assets often include governance rights, but voter turnout is typically low. Many investors lack the time or interest to deeply analyze proposals, reducing governance to a formality.

AI agents can change this. By analyzing proposal texts, assessing impacts on asset value, simulating different voting outcomes, they can make decisions on behalf of investors. They can participate continuously, not just at annual meetings, turning governance into a routine activity rather than a sporadic formality.

3. The market is already voting with real money

These projections are not just forecasts; market data already confirms the trend.

RWA.xyz shows that as of March 2026, the on-chain value of tokenized real-world assets (excluding stablecoins) exceeded $25 billion, nearly quadrupling from a year earlier. The six major asset classes—U.S. Treasuries, commodities, private credit, institutional alternative funds, corporate bonds, and non-U.S. government debt—each have on-chain scales over $1 billion.

Traditional financial giants are accelerating their involvement. BlackRock launched a tokenized fund on Ethereum (BUILD), Franklin D. moved a U.S. government money market fund (FOBXX) to Solana, and JPMorgan processed billions in tokenized repo transactions via Kinexys. These institutions are unlikely to enter a market without promising prospects.

In AI infrastructure, the competition between Circle and Stripe is especially noteworthy. Both are extending into each other’s domains: Circle through Arc L1 blockchain, CCTP cross-chain transfer protocol, and Circle Payments Network; Stripe via its x402 USDC payment on Base, a $1.1 billion acquisition of Bridge, and joint development of Tempo L1 settlement chain with Paradigm.

Artemis data shows that in January, USDC on-chain trading volume exceeded $84 trillion, with the total stablecoin market surpassing $300 billion. This is a capital scale sufficient to support an AI agent economy.

Meanwhile, OpenAI and Paradigm’s collaboration on EVMbench is testing AI agents’ ability to detect, patch, and exploit smart contract vulnerabilities. Follow-up studies indicate that in EVMbench tests, AI agents can identify up to 65% of real-world bugs. Although end-to-end exploitation success rates are still below human experts, this data already attracts significant security industry attention.

4. The double-edged sword: great opportunities, many pitfalls

Major technological shifts always come with both opportunities and risks, and the integration of AI agents with RWAs is no exception.

Opportunities primarily lie in efficiency gains. AI agents can operate 24/7 without human limitations; monitor hundreds of markets simultaneously to capture fleeting arbitrage; execute complex strategies beyond human capacity. For asset managers, this means lower operational costs and expanded management scale.

New business models are emerging. “AI agent-as-a-service” platforms could become a growth driver: companies renting AI agents to manage RWA assets without building in-house tech. Niche sectors like cross-chain liquidity aggregation, automated market making, and algorithmic governance may spawn specialized agent service providers.

Global liquidity is another promising dimension. AI agents can seamlessly connect multiple chains, transferring assets across different blockchain ecosystems, breaking down current liquidity barriers caused by chain fragmentation. As agents freely traverse ecosystems like Ethereum, Solana, NEAR, the depth and breadth of RWA markets will significantly increase.

Challenges are equally significant.

Security risks are paramount. AI agents holding private keys, executing transactions, managing assets become prime targets for hackers. Key management vulnerabilities, algorithm flaws, adversarial attacks could lead to asset losses. EVMbench research shows that while AI agents excel at vulnerability detection, their success in real-world exploitation remains below expectations, indicating current tech is not yet ready for fully unattended asset management.

Regulatory compliance is another thorny issue. The legal status of AI agents remains unclear: if an agent’s erroneous decision causes asset loss, who bears responsibility? Developers? Deployers? Asset owners? Different jurisdictions have varying attitudes, and blockchain’s global accessibility complicates enforcement. In mainland China, according to the joint 42 document issued by eight authorities, RWA tokenization and related services are illegal; AI agents’ on-chain operations must strictly adhere to this red line.

Technical barriers are real obstacles. Enterprises adopting AI agent economies need both blockchain integration and AI deployment capabilities—challenging for traditional firms. Building multidisciplinary teams, choosing suitable partners, designing robust governance frameworks all require time and resources.

5. Preparing to get on board: four essential steps

For traditional companies and listed firms eyeing the emerging AI agent economy, clear strategic planning is crucial.

Step 1: Digitize assets first

AI agents manage digital assets, not physical ones. Companies should tokenize their real-world assets (receivables, equipment, property, IP) through compliant channels. For Chinese firms, this may involve exploring offshore pathways via Hong Kong or other jurisdictions within the scope of the 42 document.

Step 2: Pilot AI agent nodes

No need for full-scale deployment immediately. Select specific scenarios—such as cross-border payments, supply chain financing, investor relations—and collaborate with mature AI agent protocols to automate management. Gather experience from small pilots, evaluate results, then expand gradually.

Step 3: Cultivate multidisciplinary teams

AI agent economy demands cross-domain talent: blockchain-savvy personnel, AI deployment engineers, legal experts familiar with financial compliance. Developing or recruiting such talent is key to long-term competitiveness.

Step 4: Participate in standard-setting

The integration of AI agents and RWAs is still early-stage. Standards, governance rules, and compliance frameworks are evolving. Forward-looking enterprises should actively participate in industry discussions to shape favorable rules.

Conclusion: The dual faces of digital civilization are quietly converging

Reflecting on the two events mentioned at the start—the technological breakthrough of OpenClaw and the rapid growth of the RWA market—they seem independent but point to the same profound historical proposition.

Within the RWA research framework, AI and blockchain are the two sides of digital civilization. One represents ultimate productivity; the other, advanced production relations. As AI agents begin to autonomously manage on-chain assets, these two aspects are undergoing unprecedented deep integration. AI agents process information, execute strategies, and participate in games with extreme efficiency, while blockchain provides trusted asset registration, transparent rule enforcement, and trustless value transfer.

This is not merely technological stacking but an evolution of economic organization. When assets are managed autonomously by AI agents, humans will retreat to roles as rule-makers and strategy designers. What social impacts will this bring? How will governance power be distributed? How will responsibility boundaries be defined? These questions lack ready answers and require joint exploration by industry, regulators, and academia.

But one thing is certain: the on-chain economy built by AI agents quietly commenced in some version update of March 2026.