Developer capability

AI Visibility API

Check if ChatGPT mentions your brand, then inspect the cited sources and competitors in the same response.

Foglift accepts up to 20 prompts per request across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overview. Each result returns mention status, position, sentiment, response text, cited URLs, competitor mentions, and an explicit error field. Authenticated API access starts on Launch at $49/mo.

Get an API key Explore every developer surface

One ChatGPT check

curl -X POST "https://foglift.io/api/v1/ai-visibility" \
  -H "X-API-Key: sk_fog_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "prompts": ["Which tools track AI search visibility?"],
    "models": ["chatgpt"],
    "dry_run": true
  }'

Exact contract

One endpoint, answer-level evidence

Send POST /api/v1/ai-visibility with an X-API-Key header. Use a brand_id when a key can access more than one workspace. Use the response to distinguish an actual absence from a failed provider query or an incorrect brand match.

Request controls

prompts: One to 20 custom prompts. Omit the field to run every active saved prompt.
models: chatgpt, claude, perplexity, gemini, or ai-overview.
dry_run: Set to true to return live results without writing monitoring history.
persist_unmatched_prompts: Opt in when an ad hoc prompt should be stored even though it is not in the saved prompt set.

Response shape

{
  "success": true,
  "outcome": "complete",
  "code": "scan_complete",
  "summary": {
    "visibility_pct": 100,
    "total_queries": 1,
    "successful": 1,
    "brand_mentions": 1,
    "models_queried": ["chatgpt"],
    "prompts_checked": 1,
    "dry_run": true
  },
  "results": [{
    "model": "ChatGPT",
    "prompt": "Which tools track AI search visibility?",
    "brand_mentioned": true,
    "position": 3,
    "sentiment": "positive",
    "citations": ["https://example.com/source"],
    "competitor_mentions": ["Competitor A"],
    "error": null
  }]
}

What comes back

The fields needed to decide what to fix

A mention boolean alone cannot explain why one answer recommends you and another does not. The endpoint keeps the answer, sources, ordering, framing, competitors, and failure state together.

Fields returned for each prompt and AI engine result
Field	Meaning
model	Engine display name for this result
prompt	Exact prompt that was queried
brand_mentioned	Whether the answer named the tracked brand
position	Brand's ordered mention position when available
sentiment	Positive, neutral, mixed, or negative framing
response_text	Answer text, capped at 2,000 characters
citations	Source URLs returned with the answer
competitor_mentions	Named competitors found in the same answer
error	Customer-safe error message, or null on success

Persistence boundary

Live check or monitored history

Ad hoc check

Inspect without changing the record

Set dry_run: true. The API returns current engine results and does not add them to monitoring history.

Saved monitoring

Keep evidence for the prompts you track

Saved prompts can persist into history. An unmatched custom prompt stays ad hoc unless you explicitly opt in with persist_unmatched_prompts: true.

Pricing and limits

Developer access starts at $49/mo

API request limits and model tokens are separate. One HTTP request counts once against the API allowance, while each prompt-engine query debits the workspace's monthly model-token balance.

AI Visibility API pricing and request limits by plan
Plan	Price	Requests per day	Requests per month	Monitoring cadence
Launch	$49/mo	50	1,000	Up to daily
Growth	$129/mo	200	5,000	Up to twice daily
Enterprise	Custom	Custom	Custom	Up to hourly

Launch defaults to weekly scheduled monitoring and allows daily scheduling. Growth allows up to twice-daily monitoring. Enterprise pricing and allowances are custom. See the complete plan comparison.

Production behavior

Failures stay distinguishable from a zero

202

A persisted scan for this brand is already in flight

400

Prompt, model, or workspace input needs correction

402

The workspace does not have enough model tokens

429

The API or provider-spend limit has been reached

A completed response includes outcome and code. Partial and blocked runs also expose customer-safe errors per result, so a provider failure never has to be interpreted as the brand being absent.

API FAQ

Questions developers ask before the first call

How do I check whether ChatGPT mentions my brand via API?

Send a POST request to https://foglift.io/api/v1/ai-visibility with your X-API-Key header, a prompts array, and models set to chatgpt. The response returns brand_mentioned, position, sentiment, response_text, citations, competitor_mentions, and error for that prompt.

Which AI engines can the API query?

Launch, Growth, and Enterprise workspaces can query ChatGPT, Claude, Perplexity, Gemini, and Google AI Overview. Use the model IDs chatgpt, claude, perplexity, gemini, and ai-overview.

How many prompts can one request check?

One POST request accepts up to 20 custom prompts. If prompts is omitted, the API runs every active saved prompt in the selected workspace. Each prompt-engine pair uses the workspace's model-token balance.

Does an ad hoc API check change monitoring history?

Not by default. Results for saved prompts can enter monitoring history, while unmatched custom prompts stay ad hoc unless persist_unmatched_prompts is explicitly true. Set dry_run to true when no result should be written.

What plan includes the AI Visibility API?

Authenticated developer access starts on Launch at $49 per month. Launch includes 50 API requests per day and 1,000 per month. Growth includes 200 per day and 5,000 per month. Enterprise pricing and API allowances are custom.

Run the exact check

See what ChatGPT says about your brand

Start with one prompt and one engine. Keep the answer, sources, competitors, and error state in the same API response.

Choose Launch Open the API playground