Agentic Web Splits: Identity Versus Capability Bets

The emerging landscape of the agentic web is bifurcating into two distinct strategic approaches for websites seeking to gain visibility with large language models (LLMs). One path focuses on establishing a clear digital identity, while the other prioritizes showcasing functional capabilities.

This division is exemplified by two proposed methods for AI interaction. The first, represented by a concept termed 'LLMs.txt', aims to provide LLMs with definitive information about who a website owner or entity is. This approach is akin to a digital resume or a verified profile, ensuring that AI systems understand the origin and nature of the content or service being accessed.

The second strategic path, illustrated by 'WebMCP' (Web Machine Capability Protocol), focuses on detailing what a website can actually do. This involves outlining the specific functions, services, or actions a site is equipped to perform, enabling AI agents to leverage these capabilities directly. This approach is designed to facilitate seamless integration and task execution between AI agents and web services.

This split suggests a future where AI agents will need to navigate two fundamental types of information to effectively interact with the web. Understanding the 'who' behind the content or service will be crucial for trust and context, while comprehending the 'what' will be essential for task completion and utility. The article "The Agentic Web Is Splitting Into Two Bets: Identity And Capability" published on Search Engine Journal highlights this evolving dynamic.