A post by @FirstSquawk on X.

Visa released its first industry benchmark for AI-driven shopping, finding that nearly half of test runs failed. The benchmark evaluates how well AI agents complete retail transactions end to end, and the roughly 50% failure rate underscores how far agentic commerce remains from reliable, mainstream deployment. The results matter because major card networks and merchants are betting on AI assistants to drive future online sales, and poor completion rates could delay adoption or shift volume back to human checkout flows. Visa has not yet disclosed the number of tests, participating merchants, or the specific failure modes behind the misses. Watch for follow-up detail on which stages of checkout break down, whether Visa sets certification or remediation requirements for AI shopping agents, and how rival networks respond with their own benchmarks.