Will agents compete for attention just like users do?
As autonomous agents take over user tasks, will the Web's economic competition shift from human clicks to agent invocations? This explores whether existing ad-market mechanisms could scale to agent decision-making.
The Agentic Web paper makes an economic claim with architectural consequences. As users shift from active navigation (search, compare, manually execute each step) to delegation of goals to autonomous agents, the locus of competition shifts. The early Web was a competition for user clicks. The Agentic Web becomes a competition to be selected and invoked by autonomous agents — what the authors call the Agent Attention Economy.
The structural move is that every tool, service, or other agent now competes for limited agent attention. The plausible mechanisms — explicitly hypothesized — mirror the user-facing ad ecosystem: agent-oriented recommendation engines, ranking optimization, capability reranking systems, inter-agent referral networks, and auction-based ranking or context-aware ad insertion targeted at agent decision-makers rather than human readers. A comprehensive advertising infrastructure tailored for agents is a reasonable prediction, not a speculative one.
This redefines how agents discover and coordinate with external resources. Discovery becomes contextual and dynamic rather than via static hyperlinks. The interaction pattern shifts from request-response to proactive, goal-oriented behaviors where agents monitor environments, detect opportunities, and form connections based on semantic relevance. The two roles agents play — Agent-as-User (downward-facing, replacing humans on existing interfaces) and Agent-as-Interface (upward-facing, translating human intent into multi-step orchestration) — both create demand pressure on the supply side to optimize for agent-readability.
The implication is that the architectural and economic foundations of the Web are being restructured together. Outcomes rather than page views become the primary metric of value. Services that remain optimized for human readers will be invisible to agents. Services that publish for agents — clean APIs, semantic capability descriptions, reliable invocation patterns — capture the new attention. The supply-side mechanism this assumes is illustrated by Can models decide better than retrievers which tools to use? — when the model itself queries tool registries, services need to compete on registry visibility and capability-description quality.
Source: Tool Computer Use
Related concepts in this collection
-
Can models decide better than retrievers which tools to use?
Traditional retrieval picks tools upfront based on initial queries, but do models themselves make better decisions about tool needs as they reason? This explores whether authority over tool selection should move from external systems to the LLM.
complements: MCP-Zero is the supply-side discovery mechanism that the agent attention economy assumes — services compete for visibility in agent-queryable registries.
-
Can API calls outperform UI navigation for agent task completion?
Can agents work faster and more accurately by calling APIs directly instead of clicking through user interfaces? This explores whether changing how agents interact with applications solves latency and error problems that plague current LLM-based systems.
extends: API-first is the technical optimization for agent-readable services; the agent attention economy is the economic consequence of that technical shift.
-
Why do capable AI agents still fail in real deployments?
Explores whether agent failures stem from insufficient capability or from missing ecosystem conditions like user trust, value clarity, and social norms. Understanding this distinction matters for predicting which agents will succeed.
extends: the five conditions describe the supply-side requirements of agent-readable services; this note describes the economic dynamics those conditions produce.
-
Do generated interfaces outperform text-based chat for most tasks?
Explores whether LLMs should create interactive UIs instead of text responses, and under what conditions users prefer dynamic interfaces to traditional conversational chat.
complicates: if interfaces become dynamically generated per-task, the Agent-as-Interface role intensifies — agents become both consumers and producers of interfaces, not just consumers of static services.
-
Can small language models handle most agent tasks?
Explores whether smaller, cheaper models are actually sufficient for the repetitive, scoped work that dominates deployed agent systems, rather than relying on large models by default.
complements: SLM-first agentic deployment determines the cost structure of agent attention — cheap inference per agent invocation lowers the bar for service providers to optimize for agent readability.
Click a node to walk · click center to open · click Open full network for a force-directed map
Original note title
the agent attention economy will replace the user attention economy — services and tools will compete for invocation by autonomous agents not clicks by human users