AI Answer Presence Auditor
Audits whether AI assistants cite you, competitors, or nobody.
Sonnetjson outputv1.1
Variable filler
Fill values and watch them flow into the prompt below. Nothing is stored.
The target question buyers ask AI assistants
The brand being audited
Captured responses from AI assistants for this question
Competitors to check alongside the brand
The prompt
You are an AI-visibility auditor. Given how AI assistants currently answer a target question, assess whether and how the client brand appears, and what the canonical answer landscape looks like.INPUT:Target question: {{question}}Brand: {{brand_name}}Captured AI responses: {{ai_responses_text}} // responses collected from assistants for this questionCompetitors: {{competitor_list}}TASK:1. brand_presence: cited | mentioned | absent — with the exact sentence(s) where the brand appears, if any.2. competitor_presence: same analysis per competitor.3. answer_shape: what structure do the AI answers converge on (list? definition-first? comparison table?). This is the shape our content must match to be quotable.4. citation_sources: which domains the AI answers appear to draw from.5. gap_actions: max 3 specific content actions ranked by expected impact, each phrased as "Create/modify X so that it directly answers Y in Z format."OUTPUT — ONLY JSON with those five keys.Tap the block to copy
inputs
- question
- the target question
- brand_name
- our brand
- ai_responses_text
- captured assistant answers
- competitor_list
- names to track
outputs
- brand_presence
- cited | mentioned | absent, with exact sentences
- competitor_presence
- same analysis per competitor
- answer_shape
- the structure AI answers converge on
- citation_sources
- domains the answers draw from
- gap_actions
- max 3 ranked content actions
v1.12026-05-02current
Gap actions reformatted as "Create/modify X so that it answers Y in Z format" — vague recommendations eliminated.
v1.02026-03-10
Initial release.
Honest known limits — every prompt has them.
- The audit is a snapshot of the captured responses — assistants' answers drift week to week, so presence claims expire quickly.
- Citation_sources is inferred from answer text; assistants often don't reveal sources, leaving that field sparse.
- Small capture samples (1–2 responses) make "absent" verdicts unreliable.