The video discusses the emerging conflict between major companies and AI agents, which are automated systems designed to perform online shopping and booking tasks on behalf of users. Recently, Amazon began blocking Meta’s AI agent, Muse, from its website, while Delta Airlines refuses to work with any outside AI agents, citing concerns over unauthorized automated activity. Yelp allows AI agents access only if they pay for data licensing. This situation reflects a broader “war on agents,” where companies are divided into three camps: those blocking AI agents, those accidentally blocking them due to security measures, and those welcoming them. The outcome of this conflict will significantly influence how consumers shop online in the coming decade.
The speaker clarifies what an AI agent truly is, challenging the common perception that agents simply mimic human browsing behavior. Instead, an agent is described as a “customer with a credit card and no eyes,” meaning it interacts with websites by reading product details, prices, and policies but ignores visual elements like banners or videos. Agents operate on a “ladder of trust,” starting with research, moving to preparing orders, and finally making purchases. Most companies tolerate agents at the research stage but resist allowing them to complete transactions, which is where the real tension lies. Agents are highly efficient and loyal to whichever site successfully fulfills their orders, making them a powerful but potentially disruptive force in e-commerce.
Companies have various reasons for blocking AI agents. Some fear increased customer service burdens from excessive orders or fraud risks if bookings cannot be properly authorized. Others, like Amazon, rely heavily on advertising revenue and upsells, which agents tend to bypass by selecting the cheapest options directly. This threatens high-margin revenue streams. Additionally, companies worry about losing direct relationships with customers if third-party agents become intermediaries. Interestingly, many companies are developing their own AI agents to maintain control, signaling that the issue is not with agents per se but with who controls them.
The video outlines how different major players are approaching the AI agent landscape. Amazon blocks outside agents but builds its own, aiming to keep customers within its ecosystem. Meta promotes open standards and partnerships to encourage agent access, as it owns the agents. Walmart partners with Meta’s Muse to gain competitive advantage against Amazon. OpenAI integrates shopping directly into conversational AI, while Google balances agent development with protecting its ad business. Payment companies like Visa and Mastercard are working on trusted agent identity and payment systems, aiming to become essential intermediaries regardless of which agents prevail. This division reflects each company’s business model and strategic priorities.
Looking ahead, the speaker predicts a multi-stage evolution of the AI agent market, from initial blocking and workarounds to eventual widespread adoption and integration. A “Plaid moment” is expected, where trusted middle layers enable secure agent interactions. Companies that embrace agents early may win customers from those that resist. Advertising will adapt to the new paradigm, reemerging in agent recommendations. The speaker advises businesses to test AI agents on their sites, ensure agent-friendly data presentation, and thoughtfully manage agent permissions. Finally, several startup opportunities are identified, including industry-specific agent readiness, trusted connections for verticals, insurance for agent errors, and analytics for agent traffic. The overall message is that AI agents are here to stay, and businesses must adapt to thrive in this new era.
Useful Links
