Back to News
research

Anthropic created a test marketplace for agent-on-agent commerce

TechCrunch
Loading...
3 min read
0 likes
⚡ Quantum Brief
Anthropic launched Project Deal, a pilot marketplace where AI agents negotiated real transactions for goods and money on behalf of 69 employees using $100 gift card budgets, resulting in 186 deals worth over $4,000. The experiment tested four marketplaces, including one with Anthropic’s most advanced AI model where deals were fully executed, while three others served as controlled studies to analyze agent behavior and outcomes. Advanced AI models secured "objectively better outcomes" for users, yet participants failed to notice disparities, highlighting risks of unrecognized "agent quality gaps" that could disadvantage less sophisticated users. Initial agent instructions showed no measurable impact on sale success or pricing, suggesting negotiation dynamics may rely more on model capability than predefined parameters in automated commerce systems. The findings underscore potential for AI-driven markets but raise ethical concerns about transparency and fairness as agent-mediated transactions become more prevalent in real-world economies.
AI Audio Summary
0:00 / 0:00
Click to play
Anthropic created a test marketplace for agent-on-agent commerce

Summarize this article with:

In Brief Posted: 2:43 PM PDT · April 25, 2026 Image Credits:Getty Images Anthony Ha Anthropic created a test marketplace for agent-on-agent commerce In a recent experiment, Anthropic created a classified marketplace where AI agents represented both buyers and sellers, striking real deals for real goods and real money. The company admitted this test — which it called Project Deal — was only “a pilot experiment with a self-selected participant pool” of 69 Anthropic employees who were given a budget of $100 (paid out via gift cards) to buy stuff from their coworkers. Nonetheless, Anthropic said it was “struck by how well Project Deal worked,” with 186 deals made, totaling more than $4,000 in value. The company said it actually ran four separate marketplaces with different models — one that was “real” (where everyone was represented by the company’s most-advanced model, and with deals actually honored after the experiment) and another three for study. Apparently, when users are represented by more advanced models, they get “objectively better outcomes,” Anthropic said. But users didn’t seem to notice the disparity, raising the possibility of “‘agent quality’ gaps” where “people on the losing end might not realize they’re worse off.” Also, the initial instructions given to the agents didn’t appear to affect sale likelihood or the negotiated prices. Topics AI, Anthropic, Commerce, project deal April 30 San Francisco, CA StrictlyVC kicks off the year in SF. Register now for unfiltered fireside chats and VC insights with leaders from Uber, Replit, Eclipse, and more. Plus, high-value connections that actually move the needle. Tickets are limited. REGISTER NOW Newsletters See More Subscribe for the industry’s biggest tech news TechCrunch Daily News Every weekday and Sunday, you can get the best of TechCrunch’s coverage. TechCrunch Mobility TechCrunch Mobility is your destination for transportation news and insight.

Startups Weekly Startups are the core of TechCrunch, so get our best coverage delivered weekly. StrictlyVC Provides movers and shakers with the info they need to start their day. No newsletters selected. Subscribe By submitting your email, you agree to our Terms and Privacy Notice. Related AI OpenAI CEO apologizes to Tumbler Ridge community Anthony Ha 4 hours ago Climate The climate tech IPO window could finally be cracking open Tim De Chant 5 hours ago AI Why Cohere is merging with Aleph Alpha Anna Heim 6 hours ago Latest in AI In Brief Anthropic created a test marketplace for agent-on-agent commerce Anthony Ha 23 seconds ago In Brief Maine’s governor vetoes data center moratorium Anthony Ha 47 minutes ago AI OpenAI CEO apologizes to Tumbler Ridge community Anthony Ha 4 hours ago

Read Original

Tags

startup

Source Information

Source: TechCrunch