When cherry-picked
A hand-written list quietly favours the questions where you already win, and hides the ones where you do not.
Before the answer, the question
Everyone measures the AI answers. Almost no one maps the questions.
You cannot measure how AI sees your brand until you know what the market actually asks. Prompt Universe builds a representative, versioned map of a category's questions, controlled for intent, coverage and bias.
a Searchestra engine · immutable, versioned manifests
The hidden variable
A prompt list is not a measurement instrument.
When lopsided
Ninety percent recommendation queries, barely any informational ones. Your coverage has a shape, and no one measured it.
When drifting
The list changed between quarters, so this quarter's numbers no longer compare with the last. The baseline moved under you.
One category, sampled
Every intent, on purpose. Not the questions that flatter you.
category: project management software · market: TR
Scroll down, and the universe fills in, one intent at a time.
informational
"how does a kanban board actually work?"
informational
"what does per-seat pricing really include?"
comparison
"trello vs asana for a small design team"
comparison
"open-source alternatives to jira"
recommendation
evidence"best tool for a 10-person B2B startup"
grounded in 47 observed queries
recommendation
"which one has the easiest onboarding?"
transactional
"cheapest plan with SSO and audit logs"
transactional
evidence"free tier limits before you must pay"
grounded in 31 observed queries
informational
"is it GDPR-compliant and EU-hosted?"
Why Prompt Universe
Representative, explainable, and frozen in time.
Representative
Sampled, not guessed
Coverage is planned across intents and scenarios, and grounded in observed demand, so no corner of the category is quietly missing.
Explainable
Every query, accountable
Each question carries its intent, scenario, source and version. You can see why it is in the universe, and defend the set to anyone.
Versioned
Immutable manifests
A locked version never changes. Measurement runs reference it, so trends stay comparable and any baseline reset is disclosed, not silent.
Plans
Start with one universe, scale to a portfolio.
Sample pricing. Plans scale by universes tracked and refresh cadence.
Starter
$199/mo
- 3 active universes
- Full intent coverage
- Immutable, versioned manifests
- Quarterly refresh
Team
$599/mo
- 15 active universes
- Evidence-grounded sampling
- Coverage & provenance reports
- Monthly refresh · workspace API keys
Scale
Custom
- Unlimited universes
- Custom generation budget & priority
- Multiple workspaces & teams
- SLA with dedicated support
Questions
What is a query universe?
A representative, controlled sample of the questions a given category, market and audience ask AI, built with explicit intent coverage and locked into an immutable, versioned manifest.
How is this different from a prompt generator?
A prompt generator answers "give me 100 prompts about CRM." Prompt Universe answers "which scenarios, intents and needs should we sample, and in what proportions, to represent real demand." It is a sampling, taxonomy and coverage problem, not just text generation.
Does it run the queries or measure visibility?
No. Prompt Universe builds the question set. Running it across AI platforms and computing visibility is a separate measurement step. The two stay strictly apart by design.
Are universes versioned?
Yes. Every locked version is immutable and carries a semantic version. A measurement run references a version, so historical numbers stay comparable and baseline changes are disclosed explicitly.