> For the complete documentation index, see [llms.txt](https://docs.airops.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.airops.com/developers/api.md).

# Insights API

The AirOps Insights API: an AEO and AI citation tracking API with programmatic access to mention rate, share of voice, citations, and page performance, for custom dashboards, ETL, and reporting.

The Insights API is AirOps' answer engine optimization (AEO) and AI citation tracking API, giving you programmatic access to the same data you see in your Insights dashboard — mention rates, citations, share of voice, page performance, and more. Use it to build custom dashboards, automate reporting, or pipe AEO data into your data warehouse alongside your other marketing metrics.

All Insights endpoints live under `/public_api/brand_kits` and require a Brand Kit with AEO enabled. For authentication setup, see the [API Reference](https://docs.airops.com/api-reference).

***

## Data Model

The Brand Kit is the root resource. Everything else connects to it:

| Resource        | What it contains                                                                                                                                     |
| --------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Analytics**   | Flexible metric queries — mention rate, share of voice, citation rate, sentiment, and more — sliced by provider, competitor, topic, country, or date |
| **Web Pages**   | Your tracked pages with AEO citations, Google Search Console, and GA4 metrics combined                                                               |
| **Prompts**     | AI questions being tracked, with mention/citation rates and trends                                                                                   |
| **Citations**   | URLs cited in AI answers, with influence scoring and sentiment                                                                                       |
| **Competitors** | Domains you're monitoring for competitive analysis                                                                                                   |
| **Topics**      | Categories that organize your prompts                                                                                                                |
| **Personas**    | Audience profiles used for prompt analysis                                                                                                           |

***

## Quick Start

### 1. Find your Brand Kit

```bash
curl -X POST https://api.airops.com/public_api/brand_kits/list \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "filters": [{ "field": "aeo_enabled", "operator": "EQ", "value": true }]
  }'
```

The response includes your Brand Kit `id` — use it in all subsequent calls.

### 2. Query your AI visibility

```bash
curl -X POST https://api.airops.com/public_api/brand_kits/{brand_kit_id}/analytics \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "metrics": ["mention_rate", "share_of_voice", "citation_rate"],
    "dimensions": ["date"],
    "grain": "weekly",
    "start_date": "2026-01-01",
    "end_date": "2026-02-18"
  }'
```

### 3. Find pages that need attention

```bash
curl -X POST https://api.airops.com/public_api/brand_kits/{brand_kit_id}/web_pages/list \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "smart_filter": "losing_ai_visibility",
    "per_page": 10,
    "sort": "-citations_count"
  }'
```

Three calls and you have: your visibility trend, competitive position, and which pages need work.

***

## Endpoints

### Which endpoint do I need?

| I want to...                            | Use                                   |
| --------------------------------------- | ------------------------------------- |
| Get my overall AI visibility metrics    | Analytics                             |
| Compare against competitors over time   | Analytics with `competitor` dimension |
| Break down performance by AI provider   | Analytics with `provider` dimension   |
| Find pages losing citations or rankings | Web Pages with smart filters          |
| Get AEO + SEO + traffic data per page   | Web Pages                             |
| See which AI questions mention my brand | Prompts                               |
| See which URLs are being cited          | Citations                             |

### Analytics

`POST /public_api/brand_kits/{id}/analytics`

The most flexible endpoint. Pick any combination of metrics and dimensions to build the exact query you need.

**Metrics:**

| Metric               | Description                                       |
| -------------------- | ------------------------------------------------- |
| `mention_rate`       | % of AI answers that mention your brand (0-100)   |
| `share_of_voice`     | Your mentions / (yours + competitors') (0-100)    |
| `citation_rate`      | % of AI answers that cite your pages (0-100)      |
| `citation_share`     | Your citations / total citations (0-100)          |
| `citation_count`     | Absolute number of citations                      |
| `sentiment_score`    | Sentiment of brand mentions (0-100)               |
| `average_position`   | Mean position in AI answer lists                  |
| `answer_count`       | Total AI answers analyzed                         |
| `first_mention_rate` | % of answers where you're mentioned first (0-100) |

**Dimensions** (up to 3 per query): `date`, `provider`, `topic`, `country`, `persona`, `domain`, `competitor`, `domain_category`

**Time grain:** `daily` (max 120 days), `weekly` (max 180 days), `monthly` (max 365 days), `total` (max 365 days)

**Filters:** `providers`, `topics`, `countries`, `personas`, `brand_mentioned`

**Limits:** Max 1,000 rows per query. Default 100.

### Web Pages

`POST /public_api/brand_kits/{id}/web_pages/list`

Your tracked pages combine citation, search, traffic, and Agent Analytics data in a single response:

| Source                    | Metrics                                                                                           |
| ------------------------- | ------------------------------------------------------------------------------------------------- |
| **AEO**                   | `citations_count`, `citation_rate`, `prompts_count`                                               |
| **Google Search Console** | `clicks`, `impressions`, `ctr`, `position`                                                        |
| **Google Analytics 4**    | `traffic`, `sessions`, `engagement`, `average_session_engagement`                                 |
| **Agent Analytics**       | `user_browse` (AI Agent Visits), `ai_training` (Training Crawls), `ai_search` (Web Search Crawls) |

The Agent Analytics fields are filterable and sortable counts for the selected date range. Connect Agent Analytics first to collect this data. These fields do not include period-over-period change variants. `total_crawls` is not a field on this endpoint; use the Agent Analytics endpoint below for that calculation.

Free-tier workspaces receive **402 Payment Required** from the Web Pages endpoint.

**Smart filters** surface common opportunities:

| Filter                  | What it finds                          |
| ----------------------- | -------------------------------------- |
| `almost_page_one`       | Pages ranking #10-20                   |
| `losing_clicks`         | Declining clicks with stable rankings  |
| `rankings_slipping`     | SERP position declining over 90 days   |
| `losing_ai_visibility`  | Pages losing AI citations              |
| `citation_rate_decline` | Losing AI authority despite stable SEO |

26+ filterable fields with operators: `EQ`, `NEQ`, `GT`, `LT`, `GEQ`, `LEQ`, `CONTAINS`, `STARTS_WITH`, `IN`

### Agent Analytics

`POST /public_api/brand_kits/{id}/agent_analytics/list`

Returns daily request counts for the Brand Kit's tracked pages after you [connect Agent Analytics](/insights/onsite/traffic.md). Each row contains `date`, `platform`, `crawl_type`, and `count`. Rows with no requests are omitted.

Provide `start_date` and `end_date` to choose a period. You can filter by `platform`, `crawl_type`, or one `web_page_id`, and sort by `date` or `count` (`-count` for largest first). Pagination defaults to 25 rows and supports up to 100 per page. Read every page before calculating totals.

| Result                                 | Calculation                                                                                                    |
| -------------------------------------- | -------------------------------------------------------------------------------------------------------------- |
| Agent Analytics **Total Crawls**       | Sum `ai_search` and `ai_training` for tracked engines.                                                         |
| **AI Agent Visits**                    | Sum `user_browse` for tracked engines.                                                                         |
| **Top Crawling Engine**                | The tracked engine with the largest combined `ai_search` and `ai_training` count.                              |
| **Crawl type breakdown**               | Each of `ai_search` and `ai_training` as a percentage of their combined total for tracked engines.             |
| **Engine breakdown**                   | Group by tracked engine and sum training crawls, web search crawls, their combined total, and AI Agent Visits. |
| **Crawl activity vs. AI agent visits** | Group by date and crawl type across all returned platforms, including `other` requests.                        |
| **AI Agent Visits over time**          | Group `user_browse` by date and tracked engine.                                                                |

Tracked engines for these calculations are `chat_gpt`, `perplexity`, `gemini`, `google_ai_mode`, and `claude`. The response also includes other platforms, such as generic search crawlers.

To match the **Pages** table's **Total Crawls** for one page, filter by its `web_page_id` and sum all rows, including every platform and crawl type. This differs from the Agent Analytics summary's AI-only crawl total.

{% hint style="info" %}
The daily endpoint follows the Agent Analytics tab's date boundary: a request at exactly midnight after `end_date` can appear as a row dated the next day. Exclude rows dated after `end_date` when matching the Pages table.
{% endhint %}

### Prompts

`POST /public_api/brand_kits/{id}/prompts/list` · `GET /public_api/brand_kits/{id}/prompts/{id}`

AI questions being tracked for your brand. Each prompt includes `mention_rate`, `citation_rate`, and period-over-period trends (`mention_rate_trend`, `citation_rate_trend`).

Filter by `text`, `brand_mentioned`, `topic_id`, `prompt_volume`, `mention_rate`, `citation_rate`. Scope metrics to specific providers, countries, or personas using context filters.

### Answer intent

Answer list and detail responses expose the prompt's read-only `intent_stage`: `awareness`, `consideration`, `evaluation`, or `purchase`. It is `null` when classification is unavailable. The same field is available through MCP answer tools.

### Citations

`POST /public_api/brand_kits/{id}/citations/list` · `POST /public_api/brand_kits/{id}/citations/show`

URLs cited in AI answers, each scored with an `influence_score` (0-100) that factors in domain authority, citation frequency, and relevance. Includes `brand_sentiment` (positive/neutral/negative) and `domain_category`.

Sort by `citation_count`, `citation_share`, `citation_rate`, or `influence_score`.

For citations and domains, `prompt_type` with `EQUALS` and value `brand` or `category` describes the prompt. `brand_mentioned` with `EQ` and value `true` or `false` describes whether the citing page mentions your brand. These filters can be combined:

```json
{
  "filters": [
    { "field": "prompt_type", "operator": "EQUALS", "value": "category" },
    { "field": "brand_mentioned", "operator": "EQ", "value": true }
  ]
}
```

The former `brand_mentioned EQUALS brand/category` and `brand_referenced EQ true/false` aliases remain accepted. The explicit new filters take precedence if both forms are present. On the Prompts endpoint, `brand_mentioned EQ true/false` continues to describe the prompt text.

Citation list and detail responses include a `social` field when supported social metadata exists. It identifies the platform and contains `content` and `account` objects with metadata, metrics, and sync status. `social` is `null` when no supported social record exists; `social.account` can be `null` even when content is present. Metadata may be incomplete while collection is pending or after a failed collection.

YouTube content metadata also includes `duration_seconds`, `format`, and `sponsorship_status`. Duration can be `null`; format and sponsorship status use `unknown` when the information is unavailable.

### Observed Ads

`POST /public_api/aeo/brand_kits/{id}/ads/list` · `POST /public_api/aeo/brand_kits/{id}/ads/show`

Query ChatGPT and Google AI Mode ads AirOps observes for the Brand Kit's tracked prompts. These endpoints use the same observed-ad data as [Ad Library](/insights/ads.md#ad-library) and do not require a ChatGPT Ads account connection. They require access to the Brand Kit's Ads data.

| Endpoint | Returns                                                                                                                                                    |
| -------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **List** | One row per provider and observed creative, advertiser details, title and body, appearances, distinct answers, prompts, frequency, and observation history |
| **Show** | One provider-specific creative's details, image and destination URL, advertiser, and observation history                                                   |

Pass the list result's `observed_creative_id` as `id` and its `provider` as a single-element `providers` array to the show endpoint. This identifies an observed creative on that platform, not a resource in your connected ad account. The same creative can have separate list rows for different platforms.

Lists include both supported providers by default. Use `providers: ["chat_gpt"]` or `providers: ["google_ai_mode"]` to select one. Show requests default to ChatGPT when omitted and require exactly one provider when supplied.

List filters include `topic_id`, `prompt_id`, and `prompt_type` (`EQUALS` with `brand` or `category`). The old `brand_mentioned` prompt-type alias is still accepted. Scope requests with `start_date`, `end_date`, `providers`, `countries`, `personas`, and `tags`. The list supports pagination and field selection.

`appearances` counts ad occurrences; `answers` counts distinct collected answers containing the ad. `frequency` is the percentage of tracked answers in scope containing it. `first_seen` and `last_seen` cover the ad's observation history, and `days_active` is the inclusive span between them. The show endpoint does not return the list's windowed appearance, answer, prompt, or frequency metrics.

These observations do not establish full market coverage, advertiser spend, or targeting intent. The equivalent MCP tools are [list\_aeo\_ads and get\_aeo\_ad](/developers/tools.md#list_aeo_ads).

### Competitors, Topics, Personas

`POST /public_api/brand_kits/{id}/competitors/list` · `GET .../competitors/{id}` `POST /public_api/brand_kits/{id}/topics/list` · `GET .../topics/{id}` `POST /public_api/brand_kits/{id}/personas/list` · `GET .../personas/{id}`

Reference data for your Brand Kit configuration. Use these IDs when filtering other endpoints.

***

## Common Patterns

### Pagination

All list endpoints use offset-based pagination with a max of 100 items per page.

```json
{ "page": 1, "per_page": 100 }
```

Responses include `meta.total_count` and `meta.total_pages`. Loop until `page >= total_pages`.

### Filtering

Pass a `filters` array in the request body:

```json
{
  "filters": [
    { "field": "citation_rate", "operator": "GT", "value": 5 },
    { "field": "brand_mentioned", "operator": "EQ", "value": true }
  ]
}
```

| Type    | Operators                             |
| ------- | ------------------------------------- |
| Text    | `EQUALS`, `CONTAINS`, `STARTS_WITH`   |
| Numeric | `EQ`, `NEQ`, `GT`, `LT`, `GEQ`, `LEQ` |
| Set     | `IN`                                  |

### Sorting

Pass a `sort` string. Prefix with `-` for descending:

```json
{ "sort": "-citation_rate" }
```

### Field Selection

Pass a `fields` array to limit returned fields:

```json
{ "fields": ["url", "citations_count", "clicks", "position"] }
```

### Date Ranges

Metric endpoints accept `start_date` and `end_date` in `YYYY-MM-DD` format. Defaults to the last 30 days if omitted.

* `end_date` must be before today
* GSC and GA4 metrics have a 4-day lag

### Data Availability

Every metric response includes a `data_availability` object:

```json
{
  "data_availability": {
    "earliest_data_date": "2025-10-15",
    "latest_data_date": "2026-02-18",
    "requested_period_has_data": true
  }
}
```

Check `requested_period_has_data` before processing results.

***

## Supported AI Providers

Use these values in the `providers` filter:

| Value                | Platform           |
| -------------------- | ------------------ |
| `chat_gpt`           | ChatGPT            |
| `gemini`             | Google Gemini      |
| `perplexity`         | Perplexity         |
| `google_ai_mode`     | Google AI Mode     |
| `google_ai_overview` | Google AI Overview |
| `claude`             | Claude             |

Claude is available on Enterprise workspaces by default. Each Claude Answer uses four billing units: four tasks on a combined-task workspace or four Answer Credits on a legacy workspace. See [AEO Answers usage and billing](/your-workspace/settings/billing.md#aeo-answers-usage-and-billing).

***

## Supported Countries

AirOps accepts these 49 region codes. Pass ISO 3166-1 alpha-2 codes in the `countries` filter:

`AE` `AR` `AT` `AU` `BE` `BR` `CA` `CH` `CL` `CN` `CO` `CZ` `DE` `DK` `ES` `FI` `FR` `GB` `GR` `HR` `HU` `ID` `IE` `IL` `IN` `IS` `IT` `JP` `KE` `KR` `MX` `MY` `NG` `NL` `NO` `NZ` `PE` `PH` `PL` `PT` `SA` `SE` `SG` `TH` `TR` `TW` `US` `VN` `ZA`

Platform coverage varies by region. A region appearing in this list does not guarantee that every AI provider collects answers there. For example, Kenya is supported for ChatGPT, Google AI Overview, Google AI Mode, and Claude; other providers may be unavailable.

***

## Errors

| Status  | Meaning                                                                                                      |
| ------- | ------------------------------------------------------------------------------------------------------------ |
| **400** | Invalid filter, operator, or sort field. Check `validation_errors` in the response.                          |
| **401** | Missing or invalid API key.                                                                                  |
| **402** | The endpoint requires a paid workspace; Web Pages requests from Free workspaces return this status.          |
| **404** | Brand Kit doesn't exist or you don't have access.                                                            |
| **412** | AEO is not configured for this Brand Kit.                                                                    |
| **422** | Analytics-specific: `end_date` is today or later, date range exceeds grain limit, or more than 3 dimensions. |

***

## Use Cases

### ETL & Data Export

AEO is a new data category that most marketing stacks don't have yet. The API lets you land it in your warehouse alongside SEO, paid, and web analytics data for cross-channel analysis.

**What to export:**

| Table              | Source endpoint                          | Sync strategy                              |
| ------------------ | ---------------------------------------- | ------------------------------------------ |
| Visibility metrics | Analytics (dimensions: date, provider)   | Daily incremental — append yesterday's row |
| Competitive trends | Analytics (dimensions: date, competitor) | Weekly incremental                         |
| Page performance   | Web Pages                                | Daily snapshot — paginate all pages        |
| Citation inventory | Citations                                | Weekly snapshot — paginate all citations   |
| Prompt landscape   | Prompts                                  | Weekly snapshot — paginate all prompts     |

**Integration options:**

* **Airflow / Dagster / Prefect** — Add a DAG that calls the API, writes JSON to S3/GCS, and loads via `COPY INTO`
* **Airbyte / Fivetran** — Use the generic HTTP API source connector with POST method and offset-based pagination
* **Snowflake Snowpark** — Call the API directly from a stored procedure and schedule with Snowflake Tasks
* **AWS Lambda + Snowpipe** — Serverless function on a cron writes to S3, Snowpipe auto-ingests

**Pipeline tips:**

* `end_date` must be a past date — the API only serves finalized data, so set it to yesterday or earlier. Requests with today's date or a future date will return a `422` error.
* GSC and GA4 metrics have a **4-day lag** — data for the most recent 3 days may not be available yet
* Use `per_page=100` (the max) to minimize round trips when paginating
* Check `meta.data_availability.requested_period_has_data` to detect empty date ranges before processing
* The analytics endpoint caps at **1,000 rows** per query — for long time ranges with high-cardinality dimensions, split into windows based on the grain limit (120 days daily, 180 weekly, 365 monthly)

The Web Pages endpoint is especially valuable for ETL because it combines AEO + GSC + GA4 in a single response — saving you from stitching three separate data sources together.

### Looker Studio

We provide a ready-to-deploy Looker Studio Community Connector that visualizes your AEO data directly — no ETL, data warehouse, or coding required.

**What you get:** Three data sources in one connector:

| Data Source                 | What it shows                                                                                                              |
| --------------------------- | -------------------------------------------------------------------------------------------------------------------------- |
| **AI Visibility Analytics** | Mention rate, share of voice, citation rate, sentiment, and more — sliced by date, provider, competitor, topic, or country |
| **Page Performance**        | AEO + Google Search Console + GA4 metrics pre-joined per URL in a single data source                                       |
| **Citations**               | Citing source URLs with domain info, sentiment, influence score, and domain authority                                      |

#### Getting your API key

1. Log in to [AirOps](https://app.airops.com) and go to **Workspace Settings**
2. Find the **API Key** section
3. Copy your API key — or click **Regenerate** if you need a new one

The key authenticates as a Bearer token. Keep it confidential and never expose it in client-side code.

> **Note:** Regenerating your API key immediately invalidates the old one. If you have existing integrations using the key, update them before regenerating.

#### Connecting to Looker Studio

1. Open the [AirOps AEO Connector](https://lookerstudio.google.com/datasources/create?connectorId=AKfycbwR2EmDc6nXQkCLQNnn8EvUk_Z30Maf8wWD1sK3T13Zoe9LeeEQy3MtK6uzi_sU_1za)
2. **Authorize** the connector when prompted by Google. You may see a "Google hasn't verified this app" warning — click **Advanced** → **Go to AirOps AEO Connector** to proceed. This only appears once.
3. Enter your **AirOps API key** when asked
4. Fill in the configuration:
   * **Brand Kit ID** — your numeric Brand Kit ID (visible in the AirOps dashboard URL, e.g., `app.airops.com/brand_kits/12345`)
   * **Data Source** — choose one of the three options: AI Visibility Analytics, Page Performance, or Citations
5. Click **Connect** — Looker Studio fetches the schema automatically
6. Click **Create Report** to start building

> **Tip:** Add the connector **three times** (once per data source) to get analytics, page performance, and citations all in one report. Each data source becomes an independent source you can use in any chart.

#### Example charts

* **Time series** — Mention Rate over time (Analytics: Date + Mention Rate)
* **Bar chart** — Share of Voice by AI Provider (Analytics: AI Provider + Share of Voice)
* **Competitor comparison** — Grouped bar of Mention Rate by Competitor
* **Top pages table** — URL with Citations, GSC Clicks, GA4 Traffic side by side (Pages)
* **Scatter plot** — Citations vs. GSC Clicks to find pages winning on both AEO and SEO (Pages)
* **Top citing domains** — Bar chart of Domain by Citations (Citations)
* **Sentiment breakdown** — Pie chart of Brand Sentiment by citation count (Citations)

#### Available fields

**AI Visibility Analytics:**

| Dimensions                                    | Metrics                                                                                                      |
| --------------------------------------------- | ------------------------------------------------------------------------------------------------------------ |
| Date, AI Provider, Competitor, Topic, Country | Mention Rate, Share of Voice, Citation Rate, Citations, Sentiment, Avg Position, First Mention Rate, Answers |

**Page Performance:**

| Dimensions  | Metrics                                                                                                                          |
| ----------- | -------------------------------------------------------------------------------------------------------------------------------- |
| URL, Folder | Citations, Citation Rate, Prompts, GSC Clicks, GSC Impressions, GSC CTR, GSC Position, GA4 Traffic, GA4 Sessions, GA4 Engagement |

**Citations:**

| Dimensions                                                                | Metrics                                                                     |
| ------------------------------------------------------------------------- | --------------------------------------------------------------------------- |
| URL, Domain, Domain Category, Page Type, Brand Mentioned, Brand Sentiment | Citations, Citation Share, Citation Rate, Influence Score, Domain Authority |

#### Dashboard layout suggestion

A single-page executive overview might include:

* **Top row** — Scorecard widgets for Mention Rate, Share of Voice, Citation Rate, Total Citations
* **Middle left** — Mention Rate over time (line chart)
* **Middle right** — Share of Voice by AI Provider (bar chart)
* **Bottom left** — Top Pages table (URL, Citations, GSC Clicks, GA4 Traffic)
* **Bottom right** — Top Citing Domains (bar chart)

Add a **Date range control** widget to the report — all three data sources respect it automatically. Data for today is excluded since it's incomplete.

#### Limitations

* The Analytics API allows a maximum of **3 dimensions** per query
* Page Performance and Citations return the **top 500 rows** per load
* This is a Community Connector — each user must authorize it on first use

### Custom Dashboards

Build internal dashboards in Retool, Metabase, or Grafana. The analytics endpoint is flexible enough to power an entire dashboard from a single endpoint — just change the dimensions.

### Automated Reporting

Schedule a script to pull weekly metrics and send a Slack summary or email digest to your team. The analytics endpoint with `competitor` and `date` dimensions gives you a full competitive report in one call.

### Alerting

Use smart filters on a schedule to detect when pages start losing AI visibility or rankings slip. Trigger alerts in Slack, PagerDuty, or email when the `losing_ai_visibility` filter returns results.

### Content Prioritization

Combine `citation_rate`, `influence_score`, and smart filters to programmatically rank which pages to optimize first. Feed the prioritized list into your content calendar or project management tool.

***

For full endpoint specifications including request/response schemas, see the [API Reference](https://docs.airops.com/api-reference/api-reference/brand-kits).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.airops.com/developers/api.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.