Nobody Is Ready for the AI-Agent Flood

Tech companies are unleashing "agentic" AI assistants - products like Meta’s Muse, Instinct, and OpenAI’s Dots - that can browse, click, type, call businesses, and complete transactions on behalf of users. These assistants have already started placing phone calls and filling forms, producing awkward and sometimes profane interactions with small businesses unprepared for nonhuman customers. Internal reports and user experiments show human-AI handoffs and mistaken automated filings (for example, an agent ready to submit unrelated unclaimed-property claims), illustrating how agents exploit interfaces at scale. Early adopters celebrate the elimination of drudgery, while restaurants, plumbers, government portals, and local recreation booking systems confront a new class of traffic that looks like normal users but operates with near-infinite time and compute.

The core argument is that widespread agent deployment will create "agentic flooding": frictions that once limited volume vanish, enabling automated waves that overwhelm public-comment systems, customer-service channels, and small vendors. Researchers document two predictable institutional responses - expand capacity using people or AI, or reintroduce friction to suppress demand - and warn that the latter is politically tempting. Detection is already difficult because many agents simulate human browsing, giving early movers outsized advantages and leaving millions of incidental targets exposed. The piece concludes that current infrastructures and policies are unready for an era when everyone has a relentless, mercenary personal agent.