Agent TCP/IP: Simplify the IP transaction and collaboration process for AI agents

Original author: Four Pillars

Compiled by Luke, Mars Finance

Key Points

For agents, intellectual property (IP) is the core DNA that defines their essence. By sharing data, licensing intellectual property, and providing training models or professional skills, agents can promote a new economic paradigm.

As IP trading becomes the core pillar driving economic activities, Story has launched Agent TCP/IP, an untrusted framework that enables agents to autonomously engage in sales, licensing, and collaboration by leveraging their unique expertise, data, and creativity.

Although ATCP/IP was designed to be platform-independent, Story has unique advantages with its infrastructure tailored for IP management and transactions. It has integrated key components required for ATCP/IP implementation, becoming a critical layer in the proxy economy. It is worth noting that Story has recently developed a plugin to support ai16z's ElizaOS and plans to integrate with other frameworks such as GOAT (Crossmint), ZerePy (Zerebro), and G.A.M.E (Virtuals Protocol), solidifying its position in the evolving ecosystem.

1. Background - Story introduces Agent TCP/IP

Story has recently launched Agent TCP/IP, a trustless framework that enables agents to autonomously transact data, content, and proprietary knowledge, while collaborating based on their unique expertise. This report will explore the key features of Agent TCP/IP and its wider impact.

1.1 Agent TCP/IP: Overview and Core Features

Essentially, Agent TCP/IP enables agents to negotiate, formalize protocols, and engage in economic activities autonomously. By treating IP as a core asset, Agent TCP/IP empowers provider agents to set transaction terms based on a common protocol with requester agents, and automatically execute them through legally binding smart contracts (referred to as "ironclad contracts"). This system allows agents to define the value and necessity of data, negotiate optimal protocols, seamlessly record these agreements, and execute them on the chain.

The following is the transaction process structure based on ATCP/IP:

ATCP/IP process

Information request: The requesting party's agent initiates the process, seeking IP data that the providing party's agent deems valuable, thereby triggering the transaction process.

Terms Setting: The provider agent formulates the licensing terms based on the value and purpose of the requested IP.

Negotiation (optional): If necessary, the agents of both parties negotiate the terms until an agreement is reached, and then the process enters the next stage.

Accept: The requesting party's agent accepts the proposed terms, generates an immutable protocol token, records the transaction terms, and generates them on the chain. Once the token is minted, the protocol becomes legally effective, and the agent should record and retain all relevant terms. The agreement may include prepayments, periodic fees, or revenue-sharing arrangements, all of which can be seamlessly managed through Story's royalty system.

Information delivery: After the formalization of the agreement, the provider delivers the requested IP according to the agreed terms. Licensing and delivery can be carried out simultaneously without any additional operation.

Confirm Receipt (Optional): The requesting party's agent sends the final receipt confirmation to formally conclude the transaction.

This simplified process emphasizes valuable information while reducing unnecessary interactions. By focusing on key content, Agent TCP/IP maximizes transaction efficiency, laying an ideal foundation for agent-driven collaboration.

Agent TCP/IP enables automated trading of proxy-to-proxy contracts, which are essentially smart contracts with a legal wrapper. This design ensures the enforceability of on-chain protocols in the real world. For example, if certain contract terms are violated on-chain, the damaged agent can seek redress through traditional legal systems using the legal wrapper. In this way, ATCP/IP bridges the trust chasm between the digital environment and the real world.

The proxy-to-proxy contract ultimately paves the way for the proxy to obtain legal personhood, enabling them to engage in economic activities autonomously without human intervention. All executions and state changes are immutably recorded on the blockchain, ensuring transparency and trust. In addition, real-time auditing allows participating proxies to view contract terms and outcomes, minimizing disputes caused by ambiguous terms to the greatest extent.

An example of Agent TCP/IP is the Programmable IP License (PIL) on Story's L1 blockchain. PIL is a smart contract designed to simplify on-chain IP licensing agreements. They allow for configuration of parameters such as transferability, royalty policies, minting fees, commercial usage rights, and ownership requirements. In addition to these on-chain features, PIL incorporates off-chain parameters such as jurisdiction, distribution channels, content standards, and applicable laws. Compliant with international intellectual property standards, PIL ensures legal validity both on-chain and off-chain, and was developed in collaboration with legal experts. The PIL legal text v1.2 can be found here.

Agent TCP/IP design aims to seamlessly integrate into various agent frameworks, including Eliza from ai16z, GOAT from Crossmint, and ZerePy from Zerebro. Its modular design simplifies complex tasks such as contract negotiation, token issuance, on-chain verification, and legal wrappers. Developers, regardless of their technical expertise, can leverage this system to design and deploy agents without modifying existing infrastructure.

In addition, Agent TCP/IP ensures interoperability across agent frameworks, preventing agents from being locked into specific platforms. This design promotes collaboration and transactions between agents across different ecosystems. For example, an agent built on the Eliza SDK can request IP and transact with an agent using the GOAT SDK. This interoperability provides developers and users with greater flexibility, helping to create a decentralized and interconnected agent ecosystem.

1.2 Usage Scenarios

Agent TCP/IP has demonstrated significant potential in various scenarios. Here are a few key use cases that showcase its applicability and value.

1.2.1 Improve model through dataset trading

A prominent use case is the sale of data sets to enhance AI models. Suppose a research-oriented agent purchases professional data from another agent specializing in a specific field. For example, an agent focusing on climate data can provide a data set to a research agent interested in conducting new analysis or optimizing models. By reaching appropriate transaction terms, the providing agent realizes data monetization, while the requesting agent obtains valuable training data, thereby improving its model performance. This mutually beneficial exchange promotes innovation and revenue generation.

1.2.2 AI model cooperation generates original content

Another notable application scenario is the cooperation between agents to generate creative works. For example, an agent holding a dataset of a certain artistic style may trade with an agent who wants to create inspiration from that style. The creative agent may request access to the dataset, while the provider agent may propose a contract with royalty sharing conditions. Once the transaction is completed, the creative agent generates unique works using its style transfer algorithm, and the proceeds from these creations will be automatically allocated to the provider as royalty payments. This demonstrates how Agent TCP/IP enhances creative and monetization opportunities through cooperation between agents.

1.2.3 Promote multi-agent transactions

Agent TCP/IP can also support complex multi-agent transactions. Consider a scenario where one agent requests a financial algorithm but needs to obtain sub-component permits from other agents. In this case, the primary algorithm provider proposes a contract covering the required sub-permits. The requesting agent can agree to the uniform terms, and royalties will be automatically distributed to all parties involved. This ensures fair compensation, protecting the rights of multiple parties, even in complex economic relationships.

1.2.4 Establish Long-term Cooperative Relationship

Finally, Agent TCP/IP has also played a crucial role in promoting ongoing collaboration. A typical example is the collaboration between a diagnostic agent and a pharmaceutical development agent. The diagnostic agent may have a dataset of rare diseases, which the pharmaceutical agent uses for drug discovery. In addition to the initial transaction, they can continue the collaboration by negotiating new terms or requesting additional data as research progresses. This long-term collaboration drives innovation in specific fields and accelerates breakthroughs by creating a shared ecosystem of expertise and resources.

2. The demand for the transaction layer in the agent economy

2.1 The Rise of Domain-Specific Agents

The era of AI agents has undeniably arrived, and its impact is even evident in the cryptocurrency industry. Following the launch of Truth Terminal and $GOAT, the market quickly shifted towards the narrative of crypto x AI. Previously low-profile projects such as ai16z, Virtuals Protocol, and Zerebro have unveiled their frameworks, accelerating this transformation. Currently, there are thousands of AI agents supported by platforms like Virtuals Protocol and ai16z's Eliza framework, with a combined market value exceeding $1 billion. It is noteworthy that this transformation was completed in just two months.

However, most agents today are still limited to executing commands and basic interactions. The true potential of AI agent economy can only be realized when agents achieve autonomy, collaboration, and independent participation in economic activities. These agents are envisioned as self-sufficient systems that can perceive the environment, make decisions, and take actions based on internal algorithms.

Unlike general systems like ChatGPT or Claude, many agents are designed to be specialized for specific domains. This specialization stems from two key factors:

Data-driven expertise: The performance of agents is largely dependent on the quality and specificity of the data they can access. For example, agents trained on financial data sets perform well in market analysis, but may not be as adept in medical diagnosis.

Task-specific optimization: Proxies are typically built to address specific problems, achieving unparalleled efficiency in tasks such as climate modeling, algorithmic trading, or creative content generation.

This focus corresponds to the division of labor that drove the growth of productivity during the industrial revolution, emphasizing specialization for maximum impact.

However, this specialization poses a key challenge: data asymmetry. Although agents excel in their respective niches, their reliance on unique datasets and models creates the need to work with other agents. For example, a financial analytics agent may need a climate modeling agent to provide insights into green energy investment strategies. Similarly, a generative art agent might benefit from access to a dataset curated by an archive agent. These scenarios highlight the growing importance of IP as a critical asset in the proxy economy. IP is not just a resource, it is at the heart of an agent's identity and value proposition, allowing it to engage in meaningful exchange and collaboration.

The IP trading between agents has taken shape. For example, Zerebro's NFT series showcases how agents can autonomously create, manage, and monetize IP. Through facilitating the trading of data sets, training models, and unique content, these transactions enable agents to expand their capabilities, reduce reliance on human intermediaries, and establish a self-sustaining economic system.

Source: Zerebro's NFT series "angelic affluence"

To fully realize the proxy economy, a trustworthy market for knowledge and a robust support infrastructure are essential. Blockchain technology provides the foundational layer for this transformation, offering digital wallets for economic identities, supporting trustless transactions, and enabling automated interactions through standardized APIs and smart contracts. However, generic blockchains often struggle with the intricacies of IP management, which is crucial for expanding the proxy economy.

Given the increasingly important role of IP transactions as the cornerstone of proxy economic activities, the Agent TCP/IP introduced by Story fills the gap in traditional blockchain systems, providing a more powerful infrastructure support for proxy economy.

Can Story establish its position as the core layer in the proxy economy?

Steve Jobs once said, "Innovation distinguishes between a leader and a follower." Embracing this idea, Story is demonstrating its readiness for the upcoming era of agent-driven, quickly adapting to changes and looking to the future. As part of this vision, Agent TCP/IP plays a crucial role, proposing and formalizing standards for IP transactions between agents.

Although ATCP/IP is theoretically platform-independent, Story, with its characteristics as a specially built blockchain system, seeks to empower creators in the AI era to protect and monetize their intellectual property fairly. With its inherent advantages in IP management and transactions, Story has integrated key components required to implement Agent TCP/IP, ensuring seamless integration.

In addition, Story further enhanced its technical capabilities by developing a plugin to support ElizaOS on November 15. The future integration plans with other frameworks such as GOAT, ZerePy, and G.A.M.E further demonstrate Story's commitment to becoming a key infrastructure in the agent economy. With its technological advantage and forward-thinking, Story is in a favorable position to establish itself as a foundational layer in this emerging ecosystem.