The AI visibility API. Eight engines, one JSON.
How often ChatGPT, Google AI Mode, Google AI Overviews, Gemini, Claude, Perplexity, Grok and DeepSeek recommend a brand when buyers ask, where it lands in the answer, and what would move it. Priced per answer, evidence attached.
8
AI engines, each response tagged with its source engine
$0.08
per answer on every engine, held at submission, settled to what arrived
10 min
from sign-up to first results: create a key, one POST, no sales call
What it returns
Visibility as a number a client can act on.
Recommend rate is the share of answers in a task in which the brand was named as a recommendation, not merely mentioned. Each answer carries the engine, the prompt, the position, the competitors named and the sentiment, so the rate, share of voice and sentiment mix are one GROUP BY away. Because AI engines answer differently run to run, run the same task on a schedule and the answer count is in every payload.
- Per engine, per prompt. Filter by any of the eight engines, by prompt group or by a single prompt. Every row is tagged with where it came from.
- Repeated runs. Single checks are noise on this surface. Run the same task daily or weekly and compute the rate over the answers; the count is in the results.
- Bring your own prompts. Submit a prompt set and a brand as a task and get the same enriched answers back, billed per answer.
Try the API
Real answers, in the exact shape the API returns.
Pick a brand and an engine to see the call, the JSON and what it means. These are answers we captured in our own batch runs (ChatGPT on 2 Oct 2026, Claude on 29 Jul 2026, Perplexity on 29 Jul 2026), replayed as recorded responses. This page makes no live call.
Brand
Engine
What came back
“What's the best CRM for a SaaS startup with a small sales team?”
ChatGPT, captured 2 Oct 2026
- Named
- Yes, 1st of 4 brands
- Sentiment
- positive (0.88 confidence)
- Sources
- 4 cited, 1 naming HubSpot
- Cost
- $0.08 for this answer (1 prompt on 1 engine)
Also named
- Attio
- Pipedrive
- Close
Cited domains
- hubspot.com
- attio.com
- dupple.com
- close.com
Run your own prompts at $0.08 an answer.
curl -X POST https://app.honeyb.ai/api/platform/v1/tasks \
-H "Authorization: Bearer $HONEYB_API_KEY" \
-H "Content-Type: application/json" \
--data @- <<'JSON'
{
"brand": {
"name": "HubSpot",
"url": "https://hubspot.com"
},
"prompts": [
"What's the best CRM for a SaaS startup with a small sales team?"
],
"engines": [
"chatgpt"
],
"language": "en"
}
JSONRecorded responses, not live calls. Answer text, brands named and cited links are as captured; ChatGPT answers show the captured opening. In this sample, mention counts are counted in that text, sentiment is labelled by hand and page types are read from the URL. A live task returns all of them from the full answer. Task ids are illustrative, and the task is shown as the POST returns it, still running.
Endpoints
Five endpoints from the answer to the action.
Read the aggregates, or submit your own prompts as a task and get every enriched answer back.
Submit a brand, up to 300 prompts and the engines to ask. The price is held at submission and settled to the answers that arrived.
Task status per engine: queued, running, completed or failed, with the settled cost.
Every answer, enriched: brand and competitor mentions, first-mention position, sentiment with confidence, and the answer text.
Every page the engines cited, with domain, page type and position in the answer.
Your tasks, newest first, with status and spend, for reconciliation.
Build with it
Where an AI visibility API goes.
Inside a marketing platform
Serve AI visibility to your users as a native feature without building an eight-engine sampling pipeline and paying for the models yourself.
Inside agency reporting
White-label the numbers: where the client stands per engine, why, and what is being done next. The reason field comes off the endpoint.
Inside alerts
Recommend rate dropped, a competitor took the top slot, sentiment turned cautious. Route it to Slack or a ticket the day it happens.
Inside research
Run a category prompt set across engines weekly and track who wins the answer. The same task API, pointed at a market instead of a brand.
Why it holds up
Built for a surface where the same question gets different answers.
Per answer, not per snapshot
Every answer is a record with its engine, prompt and sentiment confidence, so an integration can show how sure the number is, not just the number.
Grounded engines included
ChatGPT with search, Gemini grounded, Perplexity, AI Mode and AI Overviews are captured as the buyer sees them, with the sources they used.
Evidence with every number
Aggregates link to the answers, prompts and cited sources behind them. Nothing in the payload is an unexplained index.
Pricing
Pay per answer. Nothing else.
One credit unit is one enriched answer: one prompt on one engine, at $0.08 on every engine. New accounts start with $5 of credit and no card; top-ups start at $10, credits never expire, and there is no subscription to hold. Aggregate reads on an existing workspace are included.
FAQ
Frequently asked questions.
More from the Honeyb API
AI Mentions API
LLM mentions API: every AI answer that named a brand, with the engine, prompt, position, sentiment and the sources cited. Eight engines, billed per answer.
ExploreAI Overviews API
AI Overview API: the Google AI Overview and AI Mode answer with its full text and every cited source. $0.08 per prompt via the live task API; per-keyword endpoint in early access.
ExploreSERP API
SERP API: typed Google results with organic, paid, People also ask, local and video, and the AI Overview flag. Full AI Overview text and sources from its own endpoint.
ExploreRelated on Honeyb: Free AI visibility check, Sentiment analysis, Competitive benchmarking.
Free to start
Get recommended by AI search models.
Run a free AI search visibility check, then let the Honeyb agent do the work that gets you into the answers.
ChatGPT
Claude
Gemini