Agentic Web: Weaving the Next Web with AI Agents

This challenge gives rise to the notion of the Agent Attention Economy. In a manner analogous to the early Web’s competition for user clicks, external services now compete to be selected and invoked by autonomous agents. In this paradigm, the focus shifts from the human users to the agent engaged in the execution of a sophisticated task. Every tool, service, or other agent essentially competes for limited “agent attention”. To improve visibility and invocation likelihood, these entities may adopt mechanisms such as advertising, ranking optimization, or even agent-oriented recommendation and scoring systems within the service registries.

As this competition intensifies, it is reasonable to hypothesise that a comprehensive advertising infrastructure tailored for agents will emerge, which will include agent-facing recommendation engines, capability reranking systems, inter-agent referral networks, and potentially auction-based ranking or context-aware ad insertion. This shift fundamentally redefines how agents discover and coordinate with external resources, accelerating the transformation from a human-centric to an agent-centric Web. This attention-based competition among agents may ultimately become a core mechanism for resource allocation in future Web platforms, signalling a profound restructuring of both the architectural and economic foundations of the Web.

The Web is undergoing a fundamental transformation (Petrova et al., 2025; Lù et al., 2025; Chaffer, 2025). In the traditional model, users served as active navigators: searching, comparing, and manually executing each digital step. Booking a flight, for instance, required visiting multiple travel websites, comparing ticket options, checking loyalty programs, and handling confirmation emails across services. With the rise of intelligent agents, this burden is shifting. Users now increasingly delegate goals rather than execute tasks. A travel agent AI can autonomously search for optimal flights based on personal calendar availability, loyalty points, and real-time pricing. It can coordinate with hotel agents or even adjust travel plans based on weather forecasts or meeting changes (Monica, 2024; Genspark, 2025). This represents a shift from user-driven web navigation to intent-driven orchestration, where outcomes rather than page views become the primary metric of value.

3.2.1 Evolving Interaction Patterns

The Agentic Web transforms how interaction occurs in digital environments. Traditional web use follows a request-response model, where users initiate actions, retrieve data, and evaluate results manually. Agents, by contrast, engage in proactive behaviors. They discover relevant resources, identify capabilities, and form dynamic connections based on semantic relevance rather than static hyperlinks (Tupe and Thube, 2025; Sapkota et al., 2025; Acharya et al., 2025).

This change supports continuous and goal-oriented interaction across services. Agents monitor the digital environment, detect opportunities, and collaborate with other systems to fulfill objectives. Instead of navigating predefined pathways, they identify and access web resources through contextual understanding and adaptive negotiation, resulting in more responsive and flexible connectivity.

• Agent-as-User (Downward-facing): AI agents operate as autonomous web users who can independently navigate, interact with, and consume web resources (Nakano et al., 2022; Deng et al., 2023; Zhou et al., 2023b; Gur et al., 2024; OpenAI, 2025; Monica, 2024). In this role, agents replace or augment human users in web navigation and task execution, engaging with existing web interfaces and services designed for human consumption. This enables continuous, 24/7 operation for tasks such as market research, data collection, or transaction processing.

• Agent-as-Interface (Upward-facing): AI agents serve as intelligent intermediaries between human users and web systems, translating high-level user intentions into executable actions (Corporation, 2025; Thurrott, 2024; Opera, 2025; Wiggers, 2025). These agents process natural language commands from users and orchestrate complex multi-step workflows across various web services. This perspective emphasizes the agent’s role in abstracting complexity and providing streamlined human-agent interaction.

• Intelligence Dimension: What core intelligence is required for agents to function autonomously?

• Interaction Dimension: How do agents communicate and coordinate within digital ecosystems?

• Economic Dimension: How do agents generate and exchange value at scale?