> ## Documentation Index
> Fetch the complete documentation index at: https://docs.helloatlas.in/llms.txt
> Use this file to discover all available pages before exploring further.
# Bulk Processing
> Upload and extract data from large volumes of documents asynchronously using the Atlas Bulk Service API, with paginated result retrieval.
The Atlas Bulk Service API is designed for high-throughput, asynchronous document processing. Rather than tying uploads to a specific loan application or flow, you submit a batch of documents and retrieve extracted data when processing is complete. This makes it well-suited for back-office operations such as re-processing existing document archives or running nightly extraction jobs.
## When to use bulk vs. the flow API
| Scenario | Recommended API |
| ---------------------------------------------------------------- | ----------------------------------------- |
| Real-time per-application verification with cross-checks | [Flow API](/guides/document-verification) |
| High-volume async processing, no per-application tracking needed | Bulk API — this guide |
| Callback-driven results delivery | Flow API |
| Paginated batch retrieval at your own pace | Bulk API |
Authenticate with your `Client-Id` and `Client-Secret` headers.
```bash curl theme={null}
curl -X POST https://api.helloatlas.in/v1/bulk/authtoken \
-H "Client-Id: " \
-H "Client-Secret: "
```
**Response**
```json theme={null}
{
"access_token": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9...",
"expires_in": 3600
}
```
Pass the `access_token` as the `Token` header in all subsequent requests.
Submit a `product_type` and an array of file URLs. Each file entry must include a `file_url` (a pre-signed S3 URL pointing to the document image) and a `document_id` you define for tracking purposes.
```bash curl theme={null}
curl -X POST https://api.helloatlas.in/v1/bulk/multiupload \
-H "Token: " \
-H "Content-Type: application/json" \
-d '{
"product_type": "PFL_CDL",
"file_urls": [
{
"file_url": "https://files.s3.ap-south-1.amazonaws.com/uploads/pan_card.jpg",
"document_id": "1"
},
{
"file_url": "https://files.s3.ap-south-1.amazonaws.com/uploads/aadhaar.jpg",
"document_id": "2"
},
{
"file_url": "https://files.s3.ap-south-1.amazonaws.com/uploads/gst_cert.jpg",
"document_id": "3"
}
]
}'
```
**Response** — one entry per uploaded file, all sharing the same `batch_id`
```json theme={null}
[
{
"batch_id": "BATCH-30732-SIYKJ-20251121113228",
"document_id": "1",
"file_uuid": "03f03421-c933-46fc-a30c-ffb3ea8afd1d"
},
{
"batch_id": "BATCH-30732-SIYKJ-20251121113228",
"document_id": "2",
"file_uuid": "03f03422-c933-46fc-a30c-ffb3ea8afd1d"
},
{
"batch_id": "BATCH-30732-SIYKJ-20251121113228",
"document_id": "3",
"file_uuid": "03f03423-c933-46fc-a30c-ffb3ea8afd1d"
}
]
```
Save the `batch_id` (e.g. `BATCH-30732-SIYKJ-20251121113228`). You will need it to retrieve extraction results. The `document_id` in each entry matches the value you supplied in the request, letting you map results back to your internal records.
Atlas automatically detects the document type from the image content. You do not need to declare the type per file — the extracted `document_type` field in the response will tell you what Atlas identified.
Fetch processed data for all documents in a batch. Results are paginated.
```bash curl theme={null}
curl -X GET "https://api.helloatlas.in/v1/bulk/extracts?batch_id=BATCH-30732-SIYKJ-20251121113228&page=1&per_page=50" \
-H "Token: "
```
**Query parameters**
| Parameter | Default | Description |
| ---------- | ------- | ---------------------------------------------------- |
| `batch_id` | — | Required. The batch ID returned from the upload step |
| `page` | `1` | Page number (1-indexed) |
| `per_page` | `50` | Number of records per page |
**Response**
```json theme={null}
{
"data": [
{
"document_id": "1",
"file_data": {
"document_type": "PAN",
"name_as_per_pan": { "value": "D MANIKANDAN DURAISAMY", "confidence_score": 0.95 },
"pan_number": { "value": "PLZPM5601F", "confidence_score": 0.99 },
"dob": { "value": "16/07/1986", "confidence_score": 0.98 },
"document_quality_flag": { "value": "original", "confidence_score": 0.97 }
}
},
{
"document_id": "2",
"file_data": {
"document_type": "AADHAAR",
"name": { "value": "D MANIKANDAN DURAISAMY", "confidence_score": 0.95 },
"aadhaar_number": { "value": "1234 5678 9012", "confidence_score": 0.99 },
"address": { "value": "4-CH-64, NEAR COMMUNITY HALL, BHILWARA, Rajasthan, 311001", "confidence_score": 0.95 },
"date_of_birth": { "value": "16/07/1986", "confidence_score": 0.98 }
}
}
],
"page": 1,
"per_page": 50,
"total_pages": 1,
"has_next": false,
"has_prev": false
}
```
**Pagination fields**
| Field | Description |
| ------------- | ------------------------------------ |
| `page` | Current page number |
| `per_page` | Records returned on this page |
| `total_pages` | Total number of pages available |
| `has_next` | `true` if there is a subsequent page |
| `has_prev` | `true` if there is a previous page |
Iterate through pages by incrementing the `page` parameter until `has_next` is `false`.
## Document type auto-detection
You do not need to specify the document type when uploading. Atlas infers the type from the image content and sets `document_type` in the response accordingly. Supported types include `PAN`, `AADHAAR`, `BANK_STATEMENT`, `GST_CERT`, `INVOICE`, `DELIVERY_ORDER`, `DRIVING_LICENSE`, `VOTER_ID`, `SHOP_ACT_LICENSE`, `SALE_DEED`, and many others.
If Atlas cannot identify the document type, it returns an `UNRECOGNISED` error for that file. Ensure document images are legible and not obscured.