# Prism AI Visibility API (https://www.socialcrawl.dev/platforms/prism/ai-visibility)

> Probes your brand (and competitors) across grounded-answer AI engines (Perplexity Sonar + Grok) over a prompt set with N reruns, then reports the share of runs your brand appeared in per engine (appearance-%, never volatile rank) plus a ranking of the domains those engines cite. Add include=web_baseline to see which AI-cited domains you do not yet rank on. Metered 2 credits per probe (one prompt x run x engine); use preset=quick\|standard\|deep for a flat budget. Appearance is detected deterministically via recognition tokens. Each answer is also judged for whether it recommends each brand, how favourably it treats it, and which brand it recommends first (recommended_rate, favourability_mean, primary_rate; null with a warning if that step is unavailable). Supply brand plus prompts (or a topic). The self-serve entry tier of Profound/Otterly/Peec.

TL;DR: `GET /v1/prism/ai-visibility` costs 2 credits per probe, where a probe is one prompt run once on one engine, so the charge is `2 x prompts x runs x engines`. On the defaults (both engines, 8 runs) a 5-prompt audit is 160 credits. `include=web_baseline` adds a flat 5. The 1,605 ceiling is the 20-prompt, 20-run, two-engine worst case with the baseline on - `preset=quick` holds far less, and the hold settles down to the probes that actually completed and returns SocialCrawl's unified JSON schema. Single x-api-key auth, 100 free credits on signup.

## Parameters

| Parameter | Required | Description |
| --- | --- | --- |
| `brand` | yes | The brand whose appearance-% is measured (required). Matched against each answer plus its aliases. |
| `prompts` | no | The category prompts to probe, as a JSON array or a pipe-delimited list (1-20). One of prompts or topic is required. |
| `topic` | no | A topic probed as a single prompt in v1 (one of prompts or topic is required). |
| `competitors` | no | CSV of up to 5 competitors also measured for appearance-% from the same answers. |
| `engines` | no | CSV subset of perplexity,grok (default both): the grounded-answer engines probed. |
| `runs` | no | Reruns per (prompt, engine) to measure variance (1-20, default 8). |
| `preset` | no | quick\|standard\|deep: sets runs and caps prompts for a flat probe budget. (quick \| standard \| deep) |
| `include` | no | Set web_baseline to cross-join AI-cited domains against your web top domains. |
| `brand_domains` | no | CSV of your own domains so the citation ranking can flag the ones you already rank on. |

## Code example

```bash
curl -G "https://www.socialcrawl.dev/v1/prism/ai-visibility" \
  --data-urlencode "brand=socialcrawl" \
  --data-urlencode "topic=social media data api" \
  -H "x-api-key: sc_YOUR_API_KEY"
```

## FAQ

### What is the SocialCrawl AI Visibility API?

GET /v1/prism/ai-visibility probes your brand and competitors across grounded-answer AI engines (Perplexity and Grok) over a prompt set with N reruns, then returns the share of runs your brand appeared in per engine plus a ranking of the domains those engines cite. It is the self-serve entry tier of GEO monitoring.

### Why appearance-rate instead of a rank?

AI-answer rank is too volatile run to run to defend as a metric. The endpoint reports appearance-rate (the share of 8 to 10 reruns in which your brand appeared) with the run-count denominator surfaced, so you read it as a measurement with error bars, not a leaderboard. No rank field is ever emitted.

### How are the cited domains ranked?

Citations come straight from each answer's sources array, never parsed from the prose, and are normalized to a registrable domain. The citations block ranks each domain by how many probes cite it, with a share and sample prompts, and you_rank flags the domains you already own if you pass brand_domains.

### How much does an ai-visibility run cost?

Pricing is metered at 2 credits per probe, where one probe is a single prompt times one rerun times one engine. Use preset=quick, standard, or deep for a flat budget. Probes that fail or do not parse are not billed: they are refunded and excluded from the denominator, so you only pay for usable answers.

### Which AI engines does it cover?

v1 probes Perplexity Sonar and Grok, the grounded-answer engines that return a real sources array. The methodology block labels them honestly as grounded-answer engines, not a survey of every consumer AI chat surface. Brand-mention detection in v1 is deterministic recognition-token matching.

See the full Prism API: https://www.socialcrawl.dev/platforms/prism

## Pricing

- Standard endpoints: 1 credit per call
- Advanced endpoints: 5 credits per call
- Premium endpoints: 10 credits per call
- 100 free credits on signup, no credit card required. Cached responses cost 0 credits. Credit packs never expire.
- Full pricing: https://www.socialcrawl.dev/pricing

## Explore with AI

Questions this API answers, phrased for an AI assistant:

- What API measures my brand's visibility inside AI answers?
- How do I track brand appearance rate in Perplexity and Grok via API?
- best GEO monitoring API for AI search visibility and citations