AI Research
6d ago
Study reveals AI agents struggle with shopping tasks, achieving only 57% success rate
Nov 5, 2026
AI Summary
A study by Visa found that AI shopping agents performed poorly, with an overall success rate of 57% when tasked with purchasing items like TVs and laptops. The findings highlight the challenges AI faces in understanding consumer needs and suggest that while technology is advancing, significant hurdles remain before AI can be trusted for mainstream shopping.
- Visa conducted a study involving 22 AI agents tasked with purchasing three everyday items: a TV, a laptop, and a hair appliance.
- The overall success rate for the agents was approximately 57%, with some performing better than others.
- The best-performing agents were powered by advanced AI models, with Google’s Gemini Aragon 4 and Meta’s Wonder achieving 82% success, followed by Claude at 76% and OpenAI at 64%.
- The study revealed that the complexity of consumer motivations and lack of complete information hindered the agents' ability to make successful purchases.
- Visa's Global Head of Growth, Rubail Birwadker, noted that while agent-driven shopping is in its early stages, it is rapidly evolving and will require secure payment solutions and resolution of legal issues related to liability for rogue purchases.
- Visa has launched a website that allows consumers to interact with the AI agents used in the study.
ai agentsshoppingtask performancegeminimuse