> ## Documentation Index
> Fetch the complete documentation index at: https://docs.truv.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Document Processing
> Extract structured income data from pay stubs, W-2s, and tax returns using the Document Processing demo
Verify borrower income and employment by uploading documents directly to Truv. Send pay stubs, W-2s, 1099s, and tax returns through the [Document Collections API](/api-reference/document-collections/collections_create). Truv validates, classifies, and extracts structured data — no Bridge widget required.
**Integration pattern:** [Document Processing](/developers/integration/document-processing) | **Products:** Document extraction (pay stubs, W-2s, 1099s, tax returns) | **Demo:** [Document Processing](/developers/demos/mortgage/document-processing)
***
## Get started
Clone and run the [Document Processing demo](/developers/demos/mortgage/document-processing) to upload documents, validate them, and extract structured data locally.
```bash theme={null}
git clone https://github.com/truvhq/demo-apps.git
cd demo-apps
npm install && npm start
```
Open `http://localhost:5173`, select **Mortgage > Document Processing**, and process the pre-loaded test pay stubs. You can also upload your own PDF, JPEG, or PNG files.
See [full setup instructions](/developers/quickstarts-overview#demo-apps) for ngrok and environment configuration.
The demo follows this sequence:
1. **Create a user** — Your server calls [POST /v1/users/](/api-reference/users/users_create) to create a Truv user for the borrower.
2. **Create a document collection** — Your server sends Base64-encoded files to [POST /v1/documents/collections/](/api-reference/document-collections/collections_create) with the `user_id`. Truv validates each file (size, type, readability, uniqueness) and classifies it by document type.
3. **Poll for validation** — Your server polls [GET /v1/documents/collections/{id}/](/api-reference/document-collections/collections_retrieve) until all files reach `successful` status.
4. **Finalize the collection** — Your server calls [POST /v1/documents/collections/{id}/finalize/](/api-reference/document-collections/collections_finalize_create) to trigger data extraction.
5. **Receive webhooks** — Truv sends a `task-status-updated` [webhook](/api-reference/webhooks) with `status: done` when extraction completes. The payload includes a `link_id`.
6. **Retrieve extracted data** — Fetch the income report using the `link_id` from the webhook via [GET /v1/links/{link_id}/income/report/](/api-reference/account-link-income-and-employment-reports/link_detail_reports_income).
Each step maps to a specific file in the demo. Use these as reference when building your integration.
| Step | Demo file | API reference |
| -------------------- | ---------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- |
| Collection creation | [`server/routes/upload-documents.js`](https://github.com/truvhq/demo-apps/blob/main/server/routes/upload-documents.js) | [POST /v1/documents/collections/](/api-reference/document-collections/collections_create) |
| Validation polling | [`src/demos/UploadDocuments.jsx`](https://github.com/truvhq/demo-apps/blob/main/src/demos/UploadDocuments.jsx) | [GET /v1/documents/collections/{id}/](/api-reference/document-collections/collections_retrieve) |
| Finalization | [`server/routes/upload-documents.js`](https://github.com/truvhq/demo-apps/blob/main/server/routes/upload-documents.js) | [POST /v1/documents/collections/{id}/finalize/](/api-reference/document-collections/collections_finalize_create) |
| Webhook verification | [`server/webhooks.js`](https://github.com/truvhq/demo-apps/blob/main/server/webhooks.js) | [Webhooks](/api-reference/webhooks) |
| Report retrieval | [`server/routes/upload-documents.js`](https://github.com/truvhq/demo-apps/blob/main/server/routes/upload-documents.js) | [Link income report](/api-reference/account-link-income-and-employment-reports/link_detail_reports_income) |
***
## Mortgage-specific configuration
### Supported document types
| Document | Use case | Fields extracted |
| --------------------------- | -------------------- | ----------------------------------------------- |
| Pay stubs | Income verification | Employer, gross/net pay, deductions, pay period |
| W-2 forms | Annual income | Wages, federal/state taxes, employer EIN |
| 1099 forms (NEC, MISC, DIV) | Self-employed income | Non-employee compensation, payer info |
| Tax returns (1040) | Filed income | AGI, tax liability, filing status |
### Common document combinations
**Purchase / Refinance (employed borrower):**
* 2 most recent pay stubs
* Most recent W-2
* Prior year W-2 (if available)
**Self-employed borrower:**
* 1099 forms
* 1040 tax return
* Bank statements (for asset verification)
### GSE compliance
Document-extracted data supports [Fannie Mae D1C](/mortgage/d1c-aim/fannie-mae) and [Freddie Mac AIM](/mortgage/d1c-aim/freddie-mac). Every document is checked against 50M+ verified pay stubs for fraud detection. The output report uses the same schema as direct payroll connections — no changes to your LOS integration or underwriting workflow.
### Document upload via Bridge
When document upload is enabled in the [Truv Dashboard](https://dashboard.truv.com), borrowers who can't connect their payroll provider can upload documents directly within the [Truv Bridge](/developers/sdks/overview) widget. Documents uploaded through Bridge are identified by `data_source: "docs"` in the response.
### Lenient document intake
Borrowers can upload any mix of supported files — including phone photos — and Truv classifies each one and flags unreadable scans before submission. See [intelligent intake](/developers/integration/document-processing#intelligent-intake) for how recognition and selective finalization let a borrower proceed once the required documents are present.
***
## Next steps
Complete API walkthrough for Document Collections
Full API reference for upload, validate, finalize, and retrieve
Handle fraud detection and document errors
Test with Fannie Mae and Freddie Mac report IDs