Site index for AI agents: /llms.txt. Documentation pages under /docs are also served as markdown at the same URL plus .md (e.g. /docs/geo-audit.md) or via an Accept: text/markdown request.

Keyword research asks what people type. This asks what they mean.

Find, cluster and rank the prompts buyers ask answer engines, then turn each cluster into a content brief.

Map the follow-ups where the shortlist forms

Buyers ask a question, read the answer, then ask a follow-up that narrows it. Keyword tools capture the first move and none of the rest, which is where the shortlist forms.

New comparison language appears, evaluation criteria change, and competitor names enter prompts that never had them. By the time a keyword tool registers the volume, the answer has been written by somebody else's content.

How it works

Run it before you decide what to track.

Prompt discoveryThe category, comparison, problem and solution prompts buyers use across answer engines, including phrasings no keyword tool reports.
Intent clusteringGroup prompts by objective, funnel stage, urgency, role and decision language, so a cluster maps to one piece of content rather than to twelve.
Competitor answer analysisWhere competitors appear inside those answers, how they are framed, and which proof points the engines repeat.
Entity gap detectionThe products, features, people, use cases and related concepts an engine does not currently connect to your brand.
Citation opportunity mappingThe pages, references and evidence signals that would make an answer more likely to trust you on this cluster.
Prompt-to-content briefsEach cluster resolves to a brief: explainer, comparison, landing page or refresh, with the specific question it is meant to win.

What you get

A clustered map of the prompts that decide your category.
The entities to connect to your brand next.
Ranked content briefs, each traceable to the prompt that justified it.
Objection language for sales, because a repeated prompt is a repeated objection.
What it will not do

Prompt Analysis tells you what to write. Some clusters lose because the engine has no reason to trust you, and a fifth blog post will not change that. When the gap is credibility rather than coverage, the brief says so, and the work belongs to third-party sources.

From the research

Which prompts you choose changes the measured result more than most dashboard settings. AirPulse's research cluster documents a scored, versioned selection process.

Use the prompt-selection methodology to build the first roster

Questions about Prompt Analysis

The discovery process for finding, grouping and prioritising the questions buyers ask AI engines. It turns scattered conversational queries into a structured view of answer demand.

Related features

Prompt Visibility

Where you appear, rank and get cited for the prompts your buyers ask.

Content Gaps

Questions you are relevant for but not yet named in, why, and the brief to win them.

Competitor Visibility

Your share of answer against each competitor, and the sources that move it.

See the visibility pillar

Find the prompts that decide your category.

The free audit starts from your buyers' real questions rather than a keyword export.