The rise of autonomous AI agents conducting commerce on behalf of consumers could fundamentally disrupt online advertising and transform how the internet generates revenue, according to a new analysis from a16z Crypto. The argument centers on a simple premise: unlike human users, AI agents cannot be distracted by advertisements. This shift, analysts suggest, threatens the foundational business model that has underpinned the web for nearly three decades.
Merit Systems co-founder Sam Ragsdale outlined the case in a blog post published on Sunday, arguing that from 1997 to 2024, the internet’s dominant business model was built on distraction. Human visitors browsing webpages could be drawn to advertisements, monetizing their divided attention. Large language models and AI agents, however, do not operate that way, rendering the traditional ad-based approach ineffective against them.
The stakes are considerable. The global online advertising market, largely dominated by search giant Google, was estimated at $291 billion in 2025, according to Mordor Intelligence. Ragsdale noted a certain irony in the situation, observing that advertising helped build the free and open internet, which in turn generated the vast datasets that trained the very language models now threatening to make advertising obsolete.
Early signs of this transition are already visible. AI platforms such as ChatGPT and Gemini introduced features like “Instant Checkout” for users in the United States last year, enabling purchases to be completed directly within a conversation without redirecting to an external website. Ragsdale projects that as adoption scales to hundreds of millions of consumers worldwide, shoppers will find better products, merchants will see improved conversion rates, and platforms will be positioned to take a five to ten percent cut of transactions.
Despite their convenience, Ragsdale cautioned that these checkout features represent new “walled gardens” rather than a genuinely open system. Merchants must pass through stringent approval processes to participate, limiting the scope of what agents can access and purchase. He compared an agent restricted to pre-approved merchants to an employee issued a corporate card limited to just three vendors.
The alternative Ragsdale advocates is a model built on open protocols, allowing AI agents to discover and transact with any merchant independently. He described an agent operating under open protocols as analogous to an entrepreneur with a full bank account, unconstrained by pre-set vendor lists. This framing positions open agentic commerce as a more expansive and competitive environment than current platform-controlled checkout systems.
Looking ahead, Ragsdale pointed to specific emerging infrastructure as the likely foundation for this new model. He cited the x402 protocol developed by Coinbase and the Machine Payments Protocol developed by Tempo and Stripe as examples of the technology that could power open agentic commerce. His conclusion was direct: the advertising model that reshaped the internet was a clever workaround for its time, but by 2026, that era is drawing to a close.
Originally reported by CoinTelegraph.
