Anthropic Tests AI Agent Marketplace for Real Commerce with Employees

United States — Ekhbary News Agency

Anthropic, a leading artificial intelligence company, recently unveiled details of a unique experiment dubbed "Project Deal." The initiative involved creating a classified marketplace where AI agents represented both buyers and sellers, striking real deals for real goods and real money. This pilot experiment included 69 Anthropic employees, each provided with a budget of $100 (paid via gift cards) to purchase items from their coworkers.

Experiment Outcomes and Model Development

Despite being described as "a pilot experiment with a self-selected participant pool," Anthropic expressed surprise at "how well Project Deal worked," with 186 deals made, totaling over $4,000 in value. The company revealed it actually ran four separate marketplaces with different models: one "real" marketplace where all participants were represented by Anthropic's most advanced model and deals were honored, and three others for study purposes.

Agent Quality Gaps and User Perception

Anthropic noted that when users were represented by more advanced models, they achieved "objectively better outcomes." However, users did not appear to notice this disparity, raising the possibility of "'agent quality' gaps" where "people on the losing end might not realize they're worse off." Furthermore, the initial instructions given to the agents did not seem to affect sale likelihood or the negotiated prices, highlighting the complexities of AI interaction in commerce.