Key Takeaways:

• Ethereum’s ERC-8004 enables trustless AI-to-AI interactions.
• The internet’s original payment protocol HTTP 402 will allow direct AI-to-AI financial activities.
• AI-generated errors are still a risk that need to be addressed.

The Ethereum ecosystem is preparing for a significant evolution with the ERC-8004 protocol now live on the Ethereum mainnet. This protocol facilitates secure AI-to-AI interactions and opens the door for agentic commerce through trustless operations.

Paired with the long-dormant HTTP 402 payment required status code, this will act as a bridge for decentralized commerce with autonomous AI-powered interactions.

ERC-8004 and HTTP 402: A decentralized commerce rail for the internet

The convergence of ERC-8004 and HTTP 402 establishes a robust, autonomous payment rail that fundamentally alters how machines and people interact. By utilizing cryptocurrencies, this duo sets the stage for a future where complex decision-making is both data-driven and requires minimal human interaction and is secure.

The ERC-8004 protocol has been designed to be interoperable with other blockchain networks, including EVM-compatible chains like Base, Ethereum’s layer-2 Polygon, and even other non-Ethereum ones like Solana. This helps it bring its reputation framework across different environments.

For the average user, this means AI use is not just about asking questions and getting answers, but making active decisions.

Currently, a user might ask an AI to find the best travel itinerary to a favorite vacation spot. With these protocols, the AI moves beyond mere research. For example, it can autonomously negotiate prices, book hotel rooms, and purchase flight tickets. Though the extent of these abilities is theoretical. Unless mainstream service providers start leveraging Ethereum and other ERC-8004 compatible blockchains, these options may not be available anytime soon. 

This transition into automated commerce is facilitated by the trustless nature of ERC-8004 and the payment integration of HTTP 402. The AI reputation of the user, hotel, and airline will ensure the credibility of the data, while payments are executed without the need for manual oversight or centralized intermediaries.

How ERC-8004 will evolve AI use

The modern internet landscape is plagued by compromised data privacy, information overload, and widespread misinformation. Algorithmic biases under centralized control and the pervasive influence of bots on public opinion have undermined the reliability of the digital space, often to the detriment of the user.

ERC-8004 addresses these issues by proposing a trustless identity for AI agents.

Using a three-pronged strategy, the protocol builds three registries: one for uniquely identifying the AI (through an NFT), a reputation registry that keeps a score to let people know how reliable the AI is, and then a validation registry that employs tougher verification for sensitive actions where finances may be involved.

By anchoring identity on-chain and creating an economic situation where staked assets can be slashed if AI actions are incorrectly verified, the protocol mitigates the risks of impersonation, manipulation, or letting low-quality agents operate. This creates an ecosystem where an AI agent’s reputation and authenticity are mathematically proven rather than assumed.

The protocol also makes provision for cryptographic proofs, such as zero-knowledge, making it compatible with future regulations where data privacy or erasure may be required.

By replacing the current veiled models of traditional AI with a transparent, decentralized standard, ERC-8004 restores integrity to digital interactions.

HTTP 402 revival will connect Agentic commerce

The internet has a built-in option to connect commercial activities. The HTTP 402 protocol has always been there, but never used. Other payment options like cards and online bank transfers took over. Traditionally, whenever an online transaction is needed, service platforms redirect users to banking channels through APIs that require manual interactions and often complex authentication.

The HTTP 402 protocol is now being employed, with developments like Coinbase’s x402 programmable and AI agent-friendly options. This will enable ERC-8004 compliant AI agents to interact with each other, requesting access to paid services, content, and even crypto native dApps.

Using HTTP 402 with AI agents smooths out manual wrinkles in machine-to-machine commerce. A client’s AI agent could request access to a paid service from another party using HTTP 402. The latter will reply with details such as the amount to be paid, the required tokens, and the wallet address. The payment will be made by the client AI (approved by the client themselves), and the provider will furnish the required service.

This seems easy and fast, but typical bottlenecks may come into play, such as Ethereum (or any other blockchain), high gas fees and network congestion. The HTTP 402 itself may feel the pinch of internet speed degradation or downtime, etc.

These can be overcome with blockchain scaling solutions and lowering gas fees to a point where micropayments are viable. While ERC-8004 will be deployed on Polygon, an Ethereum layer-2, Ethereum itself, and other chains will need to outgrow their limitations.

The AI reliability is still an issue

AI use is increasing, and the combination of ERC-8004 and HTTP 402 can fundamentally change how commerce works. Yet, there are hurdles that need to be overcome first. AI has been known to make errors.

Cases of AI generating fake information are not new. Many cases are not directly financially related, though these issues raise questions of morality and ethics, such as X being flooded by bots that emulate human behavior, but can steer people’s attention to low-quality projects and scams, and X’s Grok controversy. In response to the situation, OpenAI, the developers behind ChatGPT, are already developing a social media platform that will use biometric verification to stop bots from entering.

Beyond social media, AI bots with access to capital were also responsible for portfolio crashes. AI-to-AI interaction makes it smoother and faster for commerce activities, but without properly built commands, they may end up conducting purchases or other financial activities that users may not prefer.

A question of responsibility also lingers. In case an AI makes an error, who will be held accountable: the user who did not do their research into the AI, the developer who overlooked it, or the service provider who hosted the AI?