> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryxlr8.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Core Concepts

> Key terms and concepts you need to understand how XLR8.ai works.

## Experiments

An **experiment** is the core unit of work in XLR8.ai. It defines:

* Your company and domain being analyzed
* The type of analysis (Visibility or Sentiment)
* The query categories and specific queries to test
* The AI models to test against
* Optional geo-targeting (country, region, city)

Experiments are reusable — you run the same experiment multiple times over time to track how your AI visibility changes.

### Experiment Types

<Tabs>
  <Tab title="Visibility">
    **Visibility experiments** measure how often your brand appears in AI responses.

    XLR8.ai submits queries to each selected AI model and checks whether your brand (and its aliases) are mentioned in the response. Results show:

    * **Mention rate** — percentage of queries where your brand appeared
    * **Citation presence** — whether your website was cited as a source
    * **Competitor comparison** — which other brands appeared for the same queries

    **Subtypes:**

    * `discovery` — broad discovery queries ("best project management tools")
    * `informational` — informational queries about your category
    * `competitor_alternatives` — queries asking for alternatives to competitors
  </Tab>

  <Tab title="Sentiment">
    **Sentiment experiments** measure how AI models describe and perceive your brand.

    XLR8.ai collects AI responses and analyzes the tone, framing, and comparative positioning of your brand. Results show:

    * **Brand sentiment** — positive, neutral, or negative perception
    * **Competitor comparison** — how you're positioned relative to competitors
    * **Action items** — specific improvements to shift AI sentiment

    **Subtypes:**

    * `brand_sentiment` — how AI models describe your brand overall
    * `competitor_comparison` — how AI models compare your brand to competitors
  </Tab>
</Tabs>

## Executions

An **execution** is a single run of an experiment. Each time you click **Rerun Experiment** (or a schedule triggers), a new execution is created. Creating a new experiment also starts its first execution automatically.

Executions have the following statuses:

| Status        | Description                                       |
| ------------- | ------------------------------------------------- |
| `pending`     | Execution queued, not yet started                 |
| `in_progress` | Queries are actively being submitted to AI models |
| `updating`    | Results received, analysis being computed         |
| `success`     | Execution completed successfully                  |
| `failed`      | Execution encountered an error                    |

You can run multiple executions of the same experiment over time to build trend data and compare how your visibility changes.

## AI Models

XLR8.ai supports testing against multiple AI models simultaneously. Each model represents a distinct AI search engine or assistant.

Selectable models currently exposed by the frontend include:

| Model ID                       | Display Name                   |
| ------------------------------ | ------------------------------ |
| `perplexity`                   | Perplexity                     |
| `gpt-fast-gpt-5.1`             | GPT Fast (GPT-5.1)             |
| `gpt-thinking-gpt-5.1`         | GPT Thinking (GPT-5.1)         |
| `gpt-fast-gpt-5.3-chat-latest` | GPT Fast (GPT-5.3 Chat Latest) |
| `gpt-thinking-gpt-5.4`         | GPT Thinking (GPT-5.4)         |
| `grok-4`                       | Grok 4                         |
| `grok-4.20`                    | Grok 4.20                      |
| `gemini`                       | Gemini                         |
| `claude-opus-4.8`              | Claude Opus 4.8                |
| `claude-opus-4.7`              | Claude Opus 4.7                |
| `claude-sonnet-5`              | Claude Sonnet 5                |
| `claude-sonnet-4.6`            | Claude Sonnet 4.6              |
| `claude-sonnet-4.5`            | Claude Sonnet 4.5              |
| `google_ai_mode`               | Google AI Mode                 |
| `microsoft_copilot`            | Microsoft Copilot              |

<Note>
  The New Experiment picker loads models from the platform API (`/experiments/llm-models`). Display names come from the frontend model map.
</Note>

## Category Map

A **category map** organizes your experiment queries into labeled groups. For example:

```json theme={null}
{
  "Project Management": [
    "best project management tools for remote teams",
    "top task management software 2024"
  ],
  "Competitor Alternatives": [
    "alternatives to Asana",
    "tools similar to Monday.com"
  ]
}
```

XLR8.ai's domain analyzer suggests categories and queries automatically based on your company domain and experiment type.

## Company Aliases

**Aliases** are alternative names, abbreviations, or spellings of your brand. XLR8.ai uses aliases when scanning AI responses to ensure it correctly identifies all mentions of your brand.

For example, if your company is "Acme Corporation," you might add aliases: `["Acme", "Acme Corp", "ACME"]`.

## Schedules

A **schedule** automates experiment executions on a recurring basis. Available frequencies include daily, weekly, bi-weekly, and monthly.

Each schedule is attached to a specific experiment and triggers a new execution automatically.

## Content Sections

When XLR8.ai generates content, it produces multiple **sections** — structured blocks that form the full article. Each section has:

* A **name** (section title/heading)
* **Content** (the written text)
* A **score** — a cosine similarity score indicating how well the section's content aligns with your target AEO query

Higher section scores indicate stronger alignment with the query, which tends to improve AI citation rates.

## Action Items

**Action items** are specific, prioritized recommendations generated from your experiment results. They suggest concrete changes — to your content, website, or messaging — that can improve your AI visibility or shift sentiment.

Action items can be marked as complete once you've addressed them.