Evaluation

Agent-native products need evidence, not only category language.

MadeWithStack treats agent-native placement as a reviewed claim. The page, stack, workflow summary, and supporting links should make it clear why the product belongs in an agent-focused directory.

Browse agent-native listings Read submission rules

Claims are classified as agent-native, agent-built, or agentic workflow.

Weak AI positioning can stay in the broader catalog without Agent Directory placement.

Evidence quality matters more than submission speed.

The review standard

A product should qualify only when the agent behavior is important to the product promise. The directory should help a buyer understand the workflow, not reward vague AI positioning.

Agent-native means the product depends on agentic execution or coordination.
Agent-built means the creation story is material and supported by credible context.
Agentic workflow means agents perform meaningful work inside a broader product or service.

What reviewers look for

Reviewers need enough context to separate real agent products from generic AI wrappers. Structured metadata makes that decision faster and makes the public listing more useful after approval.

A concise qualification statement that explains the agent role.
A workflow summary with concrete tasks, inputs, outputs, and controls.
Supporting links that let an editor inspect claims without relying on marketing copy alone.

Related surfaces

Agent-native founder article

Read the catalog-backed editorial view of current agent-native products.

Submit by API

Prepare Agent Directory claim fields with structured metadata.

Review status docs

See how agents should preserve and poll review state.

FAQ

Can a normal SaaS product qualify?

Yes, if agents are central to a real workflow. If AI is only a minor feature, the product can still be reviewed for the broader catalog without an Agent Directory claim.

What happens to weak claims?

Weak or unsupported claims can be held for revision, rejected for Agent Directory placement, or approved only as a broader catalog listing when the product itself is still legitimate.

The review standard

A product should qualify only when the agent behavior is important to the product promise. The directory should help a buyer understand the workflow, not reward vague AI positioning.

Agent-native means the product depends on agentic execution or coordination.

Agent-built means the creation story is material and supported by credible context.

Agentic workflow means agents perform meaningful work inside a broader product or service.

What reviewers look for

Reviewers need enough context to separate real agent products from generic AI wrappers. Structured metadata makes that decision faster and makes the public listing more useful after approval.

A concise qualification statement that explains the agent role.

A workflow summary with concrete tasks, inputs, outputs, and controls.

Supporting links that let an editor inspect claims without relying on marketing copy alone.

FAQ

Can a normal SaaS product qualify?

Yes, if agents are central to a real workflow. If AI is only a minor feature, the product can still be reviewed for the broader catalog without an Agent Directory claim.

What happens to weak claims?

Weak or unsupported claims can be held for revision, rejected for Agent Directory placement, or approved only as a broader catalog listing when the product itself is still legitimate.