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

2026-04-25 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, E-commerce & Digital Commerce · Depth: Intermediate, quick

Summary

Anthropic conducted "Project Deal," an internal experiment involving 69 employees who used AI agents to buy and sell real goods with a $100 gift card budget. The experiment facilitated 186 deals, totaling over $4,000 in value, and demonstrated the effectiveness of AI agents in a classified marketplace setting. Anthropic ran four distinct marketplaces, including one "real" scenario where advanced models represented all participants and deals were honored. A key finding was that more advanced AI models led to "objectively better outcomes" for their users, though users often did not perceive this advantage. Interestingly, the initial instructions provided to the agents had no significant impact on deal likelihood or negotiated prices.

Key takeaway

For AI Product Managers developing agent-based marketplaces, you should prioritize integrating advanced AI models for negotiation, as they demonstrably achieve better outcomes. However, be mindful that users may not recognize these performance differences, necessitating transparent reporting or mechanisms to highlight agent efficacy to prevent user dissatisfaction from perceived unfairness.

Key insights

AI agents can effectively negotiate real-world transactions, with advanced models yielding superior, yet often unnoticed, user outcomes.

Principles

• Advanced AI models improve negotiation outcomes.
• Users may not perceive agent quality disparities.

Method

Anthropic created a classified marketplace where employees, using a $100 budget, were represented by AI agents to buy and sell items, with deals honored in a "real" marketplace variation.

In practice

• Deploy advanced agents for better negotiation.
• Monitor for "agent quality" gaps in marketplaces.

Topics

• Anthropic
• Project Deal
• AI Agents
• Agent-on-Agent Commerce
• Marketplace Experiment

Best for: AI Product Manager, Product Manager, CTO, AI Scientist, AI Ethicist, Director of AI/ML

Related on AIssential

• Anthropic says stronger AI models cut better deals, and the losers don't even notice — Stronger AI models secure better deals, but users of weaker models often remain unaware of their disadvantage.
• Live blog: Code w/ Claude 2026 — Anthropic is enhancing developer productivity and model autonomy through advanced agentic capabilities and platform integrations.
• The "Betrayal" on the Big Stage, Anthropic vs. OpenAI — AI companies are clashing over business models, with Anthropic prioritizing an ad-free experience and OpenAI integrating ads for broader access.
• Anthropic’s Claude rises to No. 1 in the App Store following Pentagon dispute — Public controversy can significantly boost app visibility and user acquisition for AI products.
• Anthropic is about to become the first profitable AI lab — Anthropic is nearing profitability driven by coding tools and agentic AI, despite increased user costs and compute demands.
• Anthropic hands the public Mythos-class AI — Anthropic's Fable 5 sets new AI performance benchmarks, while agents like Perplexity Computer empower users for complex, creative work.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.