How Developers Are Solving AI's Financial Limitations with Blockchain

As artificial intelligence (AI) progresses, its partnership with blockchain technology and robotics presents exciting opportunities for automation and finance. However, one of the main challenges AI faces is the inability to handle transactions independently. To overcome this obstacle, developers are now combining AI with blockchain systems, enabling AI agents to transact safely, manage digital assets, and execute smart contracts.

During the ETH Denver 2025 conference, discussions unfolded with experts from the Coinbase Developer Platform and AI and robotics firms like OpenMind and Robonomics about the role of AI-powered humanoid robots in utilizing blockchain for economic activities.

"AI on its own does not interact with blockchain," explained Nemil Dalal, a Lead Developer at Coinbase. "To enable it to have a wallet, send money, or transfer funds independently is a significant challenge. AI also cannot autonomously set up a bank account, which limits its potential to transact effectively in various scenarios."

To fill this gap, the Coinbase Developer Platform launched an initiative named Agent Kit, which empowers AI with the ability to hold a wallet and perform any on-chain action typically possible for humans.

Dalal noted that this is one of the initial use cases for AI agents, where AI can take charge of tasks such as internet monitoring and business operations on behalf of users. "Over 2,000 developers have utilized Agent Kit, generating more than $100 million in value to date. A common application is tracking on-chain activities and deciding specific actions based on set conditions, mirroring automation principles like ‘if this, then do that.’"

As AI's impact broadens across sectors, blockchain developers are keen to delve into its applications for crypto traders and investors. Dalal did caution, however, that AI hallucinations continue to be a significant issue within the field. "It might take actions or provide false information, which becomes especially critical when real money is concerned," he added.

Dalal emphasized the need for systems that grant users more control to ensure safe handling of financial transactions by AI, reducing the risk of unexpected actions. For example, if a transaction amount is below a specified limit, the AI might proceed automatically, but larger amounts could require user approval. Additionally, the AI may need to whitelist new entities it pays before making future transactions without additional checks.

Despite these concerns, Jan Liphardt, founder of OpenMind, underscored the unique advantages that AI and robotics bring to blockchain technology. "Blockchains are tailored for machines, making them essential for integrating machines effectively with humans," Liphardt stated.

As a professor of bioengineering at Stanford University, Liphardt pointed out the technological barriers that AI-driven humanoid robots face in human-centric environments, such as financial institutions and airlines, that are not built to accommodate them easily. "Observing how intelligent machines struggle within human spaces reveals a significant identity issue. To tackle this friction, we designed Iris, a robot that operates using various cryptocurrency wallets to manage transactions smoothly," he noted.

"Iris has two wallets: one straightforward Ethereum wallet and another Coinbase wallet. She safeguards the private keys of these wallets and monitors their balances every six seconds," Liphardt explained.

Balancing Innovation with Safety

The integration of robots into human systems raises essential questions regarding their identity, financial management, and interactions with people. Crucial considerations include the rules governing these robots and how they can be modified. Liphardt remarked that decentralized ledgers provide a robust framework to tackle these challenges, already accommodating various use cases.

Moreover, blockchain technology enables transparent and unchangeable governance rules that can help address concerns surrounding humanoid robots. "We encode governance and behavior rules directly onto Ethereum," Liphardt elaborated. "When a humanoid robot initializes, it accesses these rules. If someone inquires, 'Why should we trust this computer?' we can reference the governance rules, as they are unalterable and publicly available. Essentially, we've adapted Asimov’s Three Laws of Robotics to fit onto Ethereum."

While this approach provides some level of assurance, Liphardt acknowledged that it cannot entirely mitigate all safety concerns. He suggested that smaller and more user-friendly medical robots could be perceived as less intimidating than larger units designed for hazardous environments.

For Vitaly Bulatov, a proponent at Robonomics, which is responsible for the creation of the humanoid robot Unitree G1, blockchain technology offers support that smaller robotics enterprises may lack. "Though smaller LLMs can run on robots, we foresee a need for additional computational resources for the cognitive functionalities. Our platform operates on Polkadot, linking us to extensive computing services within that ecosystem," he indicated.

Bulatov believes that for robots to be effectively integrated into the economy, they must function as independent agents capable of managing financial transactions and contractual commitments. "What we are doing is providing these robots with identities, wallets, and the ability to enter binding agreements with people," he elaborated.

Every time a transaction occurs, the system gathers relevant information, including error messages and video feeds. This data is stored securely using decentralized storage on Polkadot, ensuring task completion verification. "Ultimately, whether dealing with millions of robots or a smart city structure, we can confirm that our system effectively tracks payments and technical task fulfillment," Bulatov stated.

AI, Blockchain, Robots