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

# AI Receipt OCR: Extract Transaction Data from Documents

> Upload paper receipts, invoices, or PDF bank statements and let Snishop ERP's AI extract amounts, dates, vendors, and line items automatically.

Stop typing out every receipt by hand. Snishop ERP's AI Receipt OCR lets you snap a photo or upload a document, and the AI reads it for you — pulling out every relevant detail and pre-filling your transaction form in seconds. The OCR engine handles everything from a crumpled paper receipt to a multi-page PDF bank statement, so your books stay up to date without the manual data entry.

## Supported Input Types

You can feed the OCR engine documents in several formats:

> [card] Paper Receipt Photo
>
> Take a photo with your phone or webcam. Supports JPEG and PNG.

> [card] Scanned PDF
>
> Upload a scanned invoice or receipt as a PDF file.

> [card] Bank Statement Image or CSV
>
> Upload a bank statement — as a PDF, image, or exported CSV — to import multiple transactions at once.

## What the AI Extracts

For each document you upload, the AI identifies and extracts:

| Field | Description |
| - | - |
| **Total Amount** | The final payable amount, including any discounts |
| **VAT / Tax** | PPN or other tax amounts, broken out separately |
| **Date** | Transaction or invoice date |
| **Vendor Name** | Supplier or merchant name |
| **Line Items** | Individual products or services with quantities and unit prices |
| **Invoice / Receipt Number** | Document reference number, when present |

## How to Use Receipt OCR

**1. Go to the Finance Module**

From the main navigation, click **Finance** to open the Finance module.

**2. Click Add Transaction**

Click the **+ Add Transaction** button in the top-right corner of the Transactions page.

**3. Choose Upload Receipt**

On the Add Transaction form, select the **Upload Receipt** tab instead of filling in the fields manually.

**4. Select Your File or Take a Photo**

Click **Choose File** to upload an image or PDF from your device, or click **Take Photo** to use your camera directly in the browser.

**5. AI Extracts the Data**

Snishop ERP sends the document to the OCR engine. A loading indicator appears while the AI reads the document — this typically takes 3–8 seconds depending on document complexity.

**6. Review and Confirm**

The extracted fields are pre-filled into the transaction form. Review each field, make any corrections if needed, and click **Save Transaction** to post it to your books.

> [!NOTE]
>
> AI extraction achieves high accuracy for clearly printed or typed receipts and standard PDF invoices. **Handwritten receipts** or documents with heavy background noise may require manual correction of one or more fields. Always review the extracted data before confirming.

## Bulk Import from Bank Statements

When you have an entire month's worth of transactions to import, upload your full **bank statement** (PDF or CSV) using the Bulk Import option. The AI reads every line item in your statement and automatically:

* Parses each transaction's date, amount, and description
* Matches each transaction to the most appropriate expense or income category using smart category matching
* Flags any transactions it's uncertain about for your review

To use Bulk Import:

1. Navigate to **Finance → Transactions → Bulk Import**.
2. Upload your bank statement file.
3. Review the AI-categorized transactions in the preview table.
4. Adjust any categories as needed, then click **Import All** to post them.

> [!WARNING]
>
> Always retain your **original physical or source receipts** even after digital extraction. Indonesian tax regulations may require original documentation for audits and compliance purposes. Snishop ERP's digital records do not replace your obligation to keep source documents.

> [!TIP]
>
> **Image quality matters.** For the most accurate OCR results, photograph receipts in good lighting, hold your camera straight above the document (avoid angles), and make sure the entire receipt is within the frame. A clear, flat image can significantly improve extraction accuracy — especially for smaller text like tax numbers and line-item prices.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.