AI-agent visibility creates a commercial question for marketers: what does a website gain when software accesses its content? Cloudflare’s emerging Act 4 business puts that question alongside access controls, crawling efficiency, and potential payments. For growth leaders, the challenge is to distinguish greater discoverability from measurable business value.
Agent Access Needs a Commercial Objective
Cloudflare co-founder and CEO Matthew Prince described two distinct customer priorities during the company’s third-quarter earnings call. Some customers want to attract more AI agents; others consider those visitors a business threat and want to exclude them. Cloudflare helps serve both groups, Prince said.
That split gives marketing leaders a more useful starting point than a blanket push for AI visibility. Welcoming automated visitors makes sense only when their activity supports the site’s commercial purpose. Restricting them may be appropriate when the cost of access outweighs the benefit.
Prince illustrated the imbalance with a hypothetical dining decision. An agent could retrieve menus from 1,000 restaurants before selecting one for its user. The remaining 999 would absorb the access cost without receiving the business. His example contrasts the breadth of automated research with a person browsing just a few pages.
For demand-generation teams, the implication is straightforward: agent traffic needs its own value assessment. More requests can indicate that content is being discovered, but the restaurant example shows why discovery alone cannot establish customer acquisition or revenue contribution.
Separate the signals before assigning value
| Signal | What it can establish | What it leaves unresolved |
|---|---|---|
| An agent accesses a page | The content was retrieved | Whether retrieval produced a commercial benefit |
| A crawler receives an update signal | It has information about a page change | Whether the website earns anything from that exchange |
| A site permits or blocks a crawler | An access policy is being applied | Whether that policy improves business results |
These distinctions keep AI discoverability from becoming another traffic metric presented as proof of growth.
Access Controls Are Clearer Than the Optimization Offer
Cloudflare already provides a practical mechanism for choosing which crawlers may enter a site. AI Crawl Control is available across its plans and supports decisions about individual AI crawlers. The company’s documentation identifies Pay Per Crawl as a private-beta option.
Prince said Cloudflare helps customers attract agents and manage their visits efficiently. For customers that want agents excluded, he said the company blocks them without charging those customers for the blocked traffic.
The optimization side is less specific. Although Prince compared this work to an emerging form of SEO, he did not identify a particular Cloudflare service for optimizing sites to attract agent visits during the call.
That gap limits what marketers can conclude about execution. Access management is a defined capability. The described optimization proposition does not yet provide enough detail to establish a repeatable demand-generation program.
Teams evaluating it should therefore separate two decisions: which automated visitors are welcome, and what evidence would justify investing in greater exposure. The first can be addressed through crawler permissions. The second requires a clearer connection between retrieval and the outcomes the business wants.
Openai’s Payments Validate Efficiency, Not Publisher Returns
The most concrete commercial example Prince provided involved OpenAI paying Cloudflare for information about page updates. Those signals help its crawler avoid repeatedly retrieving content that has not changed.
Cloudflare publicly introduced the research pilot on July 8, using information from participating websites, including freshness signals and detected changes. In its Oct. 15 results posting, the company said its work with OpenAI was still examining how improved discovery might translate into business outcomes.
Prince reported that Cloudflare recognized its first Act 4 revenue in the third quarter and characterized the revenue associated with agent efficiency as significant. He supplied no Act 4 amount, however. Cloudflare’s results reported total revenue of $782.5 million, representing 39% year-over-year growth, without separately identifying that business.
Nor did Prince disclose the amount OpenAI pays or when those payments began. Whether participating websites receive a portion of the money was not addressed.
For marketers and content owners, this is the central distinction: a paid infrastructure relationship does not establish a publisher payout. OpenAI’s payment demonstrates a commercial arrangement around crawling efficiency. It does not, on the available disclosures, demonstrate that the sites supplying the underlying content have gained a new revenue stream.
Payment Tools Introduce a Separate Value Proposition
Cloudflare moved Monetization Gateway and Pay Per Use into beta on Sept. 30. Monetization Gateway enables sellers to price agent access by request. Pay Per Use compensates publishers that accept an AI company’s offer based on reported usage.
Those mechanisms address a different transaction from the OpenAI efficiency arrangement. They concern charging for access or compensating content use, rather than paying Cloudflare to help a crawler work more efficiently.
Prince also outlined a possible future in which agents make small payments to websites. He suggested that an agent revealing its purpose could receive cheaper access. In his hypothetical travel example, a request describing a family’s Hawaii vacation plans could create advertising opportunities, giving the site a reason to allow access without a fee.
That is a proposed commercial model, not an established result. Prince did not specify when such payments might become a substantial revenue source, although he envisioned Act 4 becoming Cloudflare’s largest business over a 10-year horizon.
The marketing relevance lies in the disclosed intent. A request containing useful customer context could have a different value from anonymous retrieval. Whether that context can support advertising or other returns remains an open question in the example Prince presented.
Build the Evaluation Around Outcomes and Payment Flows
Cloudflare’s disclosures support an evaluation path, but not a claim that AI-agent discovery already delivers dependable growth. Leaders can assess access policy now while treating optimization and monetization as separate propositions with different evidence requirements.
Questions to resolve before committing budget
- Access: Which crawlers should be permitted, and what commercial purpose does their access serve?
- Performance: What evidence would connect agent retrieval to a useful customer or business outcome?
- Compensation: Does the arrangement pay the infrastructure provider, the content owner, or both?
- Product readiness: Is the relevant capability available, in beta, or still described as a future possibility?
Prince expects other AI companies to pay for update signals and anticipates further discussion of Act 4 in coming quarters. Those expectations should remain separate from the arrangements already disclosed.
For growth teams, the next useful evidence would be more precise product detail, demonstrated business outcomes from improved discovery, and clarity about money reaching participating websites. Until then, crawler access can be managed as an operational choice; its contribution to demand and revenue still needs to be established.
