Explore AI product websites by AI visibility score. Check public access and look for documented inputs, outputs, supported actions and agent interfaces.
Promarks a Pro listing, in gold. Pro never changes a score or a place in the order.What Pro includes →
Through a bot’s eyes
About AI Products & Agents
Building with AI does not automatically make a website readable by AI. An agent product may have an impressive demo while its public page says little about inputs, outputs or the tasks it can perform. Another assistant trying to understand that product faces the same missing information as a human buyer.
The useful distinction is between a promise and a described capability. What does the system accept? What can it return or change? Which steps require a person? Clear answers make a product easier to compare and make the boundaries of automation visible before anyone connects an account.
A discovered MCP server card identifies an interface. We do not execute its tools, benchmark the model or assess the safety and accuracy of its output.
What to look for
Use these questions when reading the sites themselves. They are guidance for this category, not additional scoring checks.
A task with defined boundaries
Look for concrete inputs and outputs, plus an explanation of what the product cannot do. Broad claims about autonomy are difficult to evaluate.
Actions separated from answers
A product that drafts a message and one that sends it have different consequences. Public documentation should make that distinction clear.
Discoverable interfaces
Capability descriptions help readers locate an integration. Check that they lead to documentation explaining access, supported operations and limitations.
Evidence behind the demo
Examples should state enough context to understand the result. Website readability alone cannot establish the quality or repeatability of model output.
Frequently asked questions
Do AI products get a scoring advantage?+
No. Being in this category awards no points. The same readability rubric applies regardless of whether the product uses AI.
Does an MCP badge mean you tested the tools?+
No. It indicates that the scan found a server card meeting our discovery check. It does not mean we connected an account or executed its advertised operations.
Does the score measure model accuracy?+
No. It measures the website's machine-readable signals and access conditions. Model accuracy, task success and output quality require different evaluations.
Questions about scores, scans and listings in general are answered in the FAQ.
