Your data team builds the monthly marketing report in a warehouse. Your agency sends its clients a deck generated from a spreadsheet. Your CEO looks at one internal dashboard and nothing else. AI visibility lives in Mentionable, and until someone copies the numbers by hand, it lives nowhere else.
The Mentionable API removes the copy. Every number you see in the dashboard is available as rows your own tools can read, in JSON or CSV, on a schedule you choose.
What does the Mentionable API return?
The Mentionable API is a read-only REST API under /api/v1 with four endpoints, one per question you would ask of your data.
Which projects can I read? GET /api/v1/projects lists the projects your key has access to: name, website, brand, creation date.
How is my visibility moving? GET /api/v1/projects/{id}/metrics returns one row per day, AI engine and prompt category: the number of answers, mentions, citations of your website, mention rate, citation rate, average position, and the split between positive, neutral and negative sentiment.
What did each AI actually answer? GET /api/v1/projects/{id}/results returns one row per AI answer: the question, the engine, the date, whether your brand was mentioned, at which position, with which sentiment, and how many competitors were named next to you. Results are paginated, newest first, up to 1,000 rows per page.
Who takes the space instead of me? GET /api/v1/projects/{id}/competitors returns one row per day, AI engine and brand: your brand and each competitor you confirmed, with mentions, average position and share of voice.
The definitions are the ones of the dashboard. A mention rate in your warehouse is the same mention rate your team reads in Mentionable, so nobody argues about which number is right.
What does a call look like?
One URL, one key, optional filters:
curl -H "Authorization: Bearer mnt_sk_..." \
"https://mentionable.ai/api/v1/projects/{id}/metrics?from=2026-09-01&to=2026-09-30&llm=chatgpt,perplexity"
Every endpoint accepts the same filters:
| Parameter | What it does |
|---|---|
from, to |
Date range, up to 366 days per call. Last 30 days by default. |
llm |
One or several AI engines, separated by commas. |
categoryId |
One prompt category only. |
format |
json (default) or csv. |
limit, cursor |
Page size and next page, on results. |
Rows are flat on purpose. Counters like mentions and results can be summed on any dimension, by month, by engine, by category, and the rates are recalculated from them, so a monthly total matches the dashboard instead of averaging daily percentages.
Errors come back as JSON with a machine code and a readable message: missing key, plan required, project out of scope, invalid query.
Where is the documentation?
At mentionable.ai/api/v1/docs, an interactive reference where you can paste a key and try every endpoint. The same contract is published as an OpenAPI file at /api/v1/openapi.json, which most HTTP clients, code generators and no-code tools import directly.
How are API keys secured?
Keys are created in Settings, MCP by an owner or an admin of the workspace. The full key is shown once, at creation. After that, the list only shows its first characters, its scope, when it was last used and whether it is active, expired or revoked.
A key can be restricted to a list of projects. It then lists and reads those projects only, and a call on any other project is refused. A key never reads more than the person who created it: if a team member only has access to two projects, their keys cannot reach a third.
The API only reads. No endpoint creates, changes or deletes anything, so a leaked key cannot alter your tracking. Revoke it in one click and the next call is refused.
Who uses the Mentionable API?
Brands with a data team. AI visibility joins the rest of your marketing data in your warehouse or BI tool, next to traffic, paid and CRM. The weekly marketing review shows it without anyone exporting a file.
Agencies. Pull every client's metrics into the reporting stack you already sell, and give a client a key restricted to their own project if their team wants the raw data.
Anyone with a spreadsheet. format=csv and a key in the URL is enough for Google Sheets or Excel to read a live table.
If you live in Looker Studio, the Looker Studio connector reads the same data without code. If you would rather ask questions in plain language, the MCP server lets Claude, ChatGPT or Cursor read your projects with the same key.
Which plans include the API?
Every Mentionable offer, brand or agency, includes the API, the MCP server and the Looker Studio connector. It also works during the 7-day free trial, so you can wire your report before you subscribe.
Reading data through the API costs nothing from your plan. The only limit is technical: 100 requests per minute per key, shared with the MCP server. Above that, the API answers 429 with a Retry-After header.