AI Perception Testing

Beta

Applies to: Agent Intelligence testing workflows

Introduced: Perception test suite beta

Updated: Documentation v1

AI Perception Testing runs controlled prompts and context checks to estimate whether machine agents can correctly interpret page intent, entities, and action pathways.

Summary

Perception tests provide structured, repeatable checks rather than free-form audits. Results help prioritize readiness and recommendation work.

Use cases

Validate key page entities and service intent
Measure gains after campaign actions
Detect regressions in machine readability

How it works

Tests sample approved page contexts, evaluate expected slots/answers, and record pass/fail and confidence-oriented evidence for dashboard review.

Inputs

Target URLs or post IDs
Expected entity and intent slots
Prompt/test profile configuration

Outputs

Per-test result summaries
Gap categories tied to readiness sections
Recommendation hints for safe follow-up actions

Admin UI location

Available through Agent Intelligence beta tooling in SmartBlocks environments.

Uses telemetry context and trend windows from the Telemetry Engine.

Test-triggered apply workflows can be audited in Visibility Ledger.

Security

Avoid sending private data to external models; use bounded prompts and redact sensitive fields where applicable.