Generate prompts automatically

Pranas suggests prompts based on your brand, audience, and category.

Last updated: 2026-05-05

Generate prompts automatically

Prompts are the questions Pranas asks AI models on your behalf during an AI Scan. The quality and breadth of your prompt set directly determines how accurately your dashboard reflects real-world AI visibility — too few prompts and you'll see noisy metrics; the wrong mix and you'll only measure one slice of intent. The prompt generator removes most of this work by reading your brand context and proposing a balanced, deduplicated library you can edit before scanning.

How generation works

When you click Generate Prompts on the Prompt Management page, Pranas sends your brand profile (name, website, category, positioning, and registered competitors) to an AI model that produces a candidate list of natural-language questions a real user might type into ChatGPT, Gemini, or Perplexity. The output is then post-processed:

Brand substitution — any literal [brand] placeholder is replaced with your canonical brand name so the prompt reads naturally and the visibility math attributes mentions correctly.
URL stripping — generated prompts never contain your domain; this prevents biased "branded" responses that would inflate your visibility score.
Type tagging — each prompt is classified as Unaided, Aided, Comparison, Problem-Solving, or Feature/Use-Case (see "Prompt types and what they reveal").
Keyword extraction — a 1:1 keyword mapping is created for each prompt so search-volume tiers can be attached.
Deduplication — near-duplicates are merged so you don't pay credits twice for essentially the same query.

Post-processing reference

Step	What it does	Why it matters
Brand substitution	Replaces `[brand]` with your canonical brand name	Keeps prompts natural and attribution correct
URL stripping	Removes your domain from prompt text	Prevents inflated branded responses
Type tagging	Classifies as Unaided, Aided, Comparison, Problem-Solving or Feature	Enables prompt-type segmentation
Keyword extraction	Maps one keyword per prompt	Attaches search-volume tiers
Deduplication	Merges near-duplicates	Avoids paying credits twice

Recommended starting library

Brand stage	Prompts to start with	Approximate generation cost
New brand, first scan	20–40	Under 1 credit
Established brand, full coverage	40–80	Roughly 1 credit
Single-launch focus	10–15	Under 1 credit

Recommended workflow

Start with 20–40 prompts for a new brand. This gives statistically meaningful coverage across all five prompt types without burning credits.
Review the Unaided prompts carefully — these are the most important because they measure organic discovery (the AI mentions you without being asked).
Edit any prompt that doesn't match how your real customers talk. The generator is good but not perfect.
Toggle prompts off (rather than deleting them) if you want to A/B test prompt sets across scans.

When to regenerate

Regenerate your prompt library when:

You change your brand positioning or category
You add or remove competitors (the generator uses competitors as context)
A new use case or feature launches
You notice your dashboard is dominated by one prompt type (rebalance via regeneration)

Regeneration does not delete existing prompts — it appends new candidates and flags duplicates so you can choose what to keep.

What it costs

Prompt generation is a small, one-time AI call billed at the same $1 = 1 credit ratio as scans. A typical generation run for 30 prompts costs less than 1 credit. The cost is shown before you confirm.

Tips for higher-quality prompts

Fill in brand positioning (problem you solve, target customer, key differentiators) on the Brands page before generating — the model uses these fields heavily.
Add at least 3 competitors so Comparison prompts feel realistic.
Use the Trending Topics suggester to inject timely, high-intent queries that the static generator wouldn't think of.

Prompt types and what they reveal
Search volume tiers
Import & export prompts
Run your first AI Scan