ENTERPRISE AI · ANALYSIS · Agent Registry
The AI Agent Transparency Gap
What 15,408 marketplace listings reveal about what enterprises know—and don't know—about the AI agents they buy
SettleTop's Global AI Registry tracks 15,408 enterprise AI agent listings from 6,888 vendors across six enterprise marketplaces. Provenance scores average just 2.0–2.8 out of 10: listings reliably show what an agent does and how to buy it, but leave the specific model, tools, data sources, data residency and permission scope unknown.
AI agents have become remarkably easy to discover and buy. AWS, Microsoft, Google Cloud, Salesforce and ServiceNow increasingly present agents as catalog products that enterprises can search, compare, subscribe to and deploy. But the information required to evaluate those agents is not keeping pace with the buying experience.
SettleTop's Global AI Registry tracks 15,408 enterprise-focused AI agent listings from 6,888 vendors across six enterprise marketplaces. Across the scored marketplaces, the average provenance score is only 2.0–2.8 out of 10. Marketplace listings commonly identify what an agent does and how to buy it, while leaving critical details—such as the specific model, tools, data sources, data residency and permission scope—unknown.