POSTService Token

LLM Crawling API

Trigger bulk brand analysis on crawled LLM responses. This API analyzes unprocessed answers and extracts brand mentions using GPT-4.

Endpoint

POST /api/llm-crawling/analyze-all

Authentication

Pass the service token in the x-service-token header.

x-service-token: your-service-token

Request Body

JSON object with the following properties:

Parameter	Type	Required	Description
`date`	string	No	Filter by date (YYYY-MM-DD). If omitted, processes all dates.
`limit`	number	No	Max answers to process (default: 100, max: 100).

Example Request

cURL

curl -X POST https://www.enception.ai/api/llm-crawling/analyze-all \
  -H "Content-Type: application/json" \
  -H "x-service-token: your-service-token" \
  -d '{"date": "2025-01-15", "limit": 50}'

JavaScript

const response = await fetch(
  'https://www.enception.ai/api/llm-crawling/analyze-all',
  {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      'x-service-token': 'your-service-token',
    },
    body: JSON.stringify({ date: '2025-01-15', limit: 50 }),
  }
);
const data = await response.json();

Python

import requests

response = requests.post(
    'https://www.enception.ai/api/llm-crawling/analyze-all',
    headers={'x-service-token': 'your-service-token'},
    json={'date': '2025-01-15', 'limit': 50}
)
data = response.json()

Response

Field	Type	Description
`message`	string	Summary of the analysis operation.
`processed`	number	Total answers processed.
`success`	number	Answers successfully analyzed.
`skipped`	number	Answers skipped (no brand).
`errors`	number	Answers that failed.
`byPlatform`	object	Results by LLM platform.

Example Response

Success (200)

{
  "message": "Analyzed 45 answers, skipped 5",
  "processed": 50,
  "success": 45,
  "skipped": 5,
  "errors": 0,
  "byPlatform": {
    "ChatGPT": { "success": 15, "skipped": 2, "errors": 0 },
    "Perplexity": { "success": 12, "skipped": 1, "errors": 0 }
  }
}

Error (401)

{
  "error": "Unauthorized"
}

Notes

Processes answers sequentially with 200ms delay to avoid rate limiting.
Each answer is analyzed using GPT-4 to extract brand mentions.
Maximum 100 answers per request.
Answers without brand names are marked as analyzed with empty brands array.
For large-scale processing, call multiple times with date filters.