---
title: "Automate repeated steps with saved workflows"
description: "Save a sequence like OCR, extract, review and export once, then run it on any document by hand, from an agent, or when you move a file into a folder."
url: https://help.pdfbetter.com/automation/saved-workflows/
collection: "Workflows & integrations"
section: "Workflows"
updated: 2026-10-18
publisher: "PDFBetter (AI Generated Inc.)"
---

# Automate repeated steps with saved workflows

> Open **Workflows**, pick a recipe or **Start empty**, add steps, choose how it starts, then click **Create workflow**. Each step uses its usual allowance (OCR pages, AI credits), and your plan sets how many workflows you can save.

- Counts toward: OCR pages, AI credits

Do it in PDFBetter: https://pdfbetter.com/app/workflows/

A workflow is a saved list of steps that runs on one document at a time, top to bottom. You can start it by hand, let a connected agent start it, or have it start when you move a document into a folder. A **Review** step pauses the run until a person decides.

## Create a workflow

1. **Open Workflows**: In the sidebar, under **AI & automation**, click **Workflows**.
2. **Pick a starting point**: Under **Recipes**, click a recipe to load its steps, or click **Start empty**. You can also click **New workflow** at the top of the page. Nothing runs until you save the workflow and start it.
3. **Name it**: Enter a **Name** and, if you like, a **Description**.
4. **Add and order the steps**: Click **Add a step** and pick one from the list. Use the arrows to move a step up or down, or remove it. Some steps ask for a setting, such as the extraction schema or the workspace.
5. **Choose how it starts**: Under **Starts**, pick **By hand or from a connected agent**, or **When a document is moved into a folder** and choose the **Folder**.
6. **Save it**: Click **Create workflow**. It now appears under **Recipes**, with its steps listed in order.

## Steps you can add

| Step | What it does | Uses |
| --- | --- | --- |
| **Recognize text (OCR)** | Adds a searchable text layer to scanned pages and saves it as a new version. | OCR pages |
| **Extract data** | Reads the fields of one of your extraction schemas, with page evidence. | AI credits |
| **Review** | Pauses the run until a person approves, corrects or rejects. | Nothing |
| **Export** | Turns the extracted values into a JSON, CSV or Excel file. | Nothing |
| **Summarize** | Writes a summary with page citations (brief, standard or detailed). | AI credits |
| **Suggest redactions** | Flags likely personal data for you to review. It never removes anything by itself. | AI credits |
| **Send webhook** | Sends a signed event to your webhook endpoints. | Business plan |
| **Add to workspace** | Files the document in a workspace for cited questions across documents. | Nothing |

The builder checks the order before you save and tells you what to fix, for example: **Export needs an Extract data step before it**, **Export must come after the review of the extracted data**, or a step that reads text **must come after OCR**. A review step must come directly after the step it checks.

## Recipes to start from

- **Invoices: OCR → extract → review → export CSV**: recognize text on scans, pull the fields from your invoice schema, let a person check them, then export a CSV.
- **Contracts: summarize → add to workspace**: write a page-cited summary of each contract and file it in a workspace.
- **Sensitive docs: suggest redactions → review**: flag likely personal data, then stop so a person can check the suggestions in the editor before anything is removed.

The extraction step needs a schema (the fields to pull out, such as invoice number, total and due date). Create one in [Extractions](https://pdfbetter.com/app/extractions/) first if the list is empty; see [Extract data from many PDFs](https://help.pdfbetter.com/ai/extract-pdf-data-table/).

## How many workflows you can save

| Plan | Saved workflows |
| --- | --- |
| Free | 1 |
| Pro | 20 |
| Business | unlimited |

The **Recipes** header shows how many you have saved out of your allowance. When you're at the limit, delete a workflow or [upgrade](https://pdfbetter.com/pricing/) to save another. Workflows need an account; they aren't available as a guest.

> **Tip:** To pause a folder-triggered workflow without deleting it, edit it and clear **Active**. Paused workflows don't start on folder uploads.

### Does uploading a file into a folder start the workflow?

The trigger is moving a document into the folder. A new upload isn't filed in any folder, so move it into the watched folder (row menu → **Move to**) and the run starts.

### If I edit a workflow, what happens to runs already going?

Runs already in progress keep the steps they started with. Your changes apply to the next run.

### What does deleting a workflow remove?

The workflow and its run history. Documents, extractions and exported files stay where they are.

## Related articles

- [Run a workflow and approve its review steps](https://help.pdfbetter.com/automation/run-workflow-approve-review.md): On the workflow under Recipes, click Run on a document and pick a document. Progress shows under Runs; a review step shows Needs approval until you click Approve or Reject.
- [Connect Claude, Cursor and other AI agents (MCP)](https://help.pdfbetter.com/automation/connect-ai-agents-mcp.md): Open Settings → Agents (MCP), name a key, choose its permissions and click Create agent key. Copy the ready-made setup for your client. Each tool call counts as one agent & automation run.
- [Get webhooks when documents are signed or forms submitted](https://help.pdfbetter.com/automation/webhook-endpoints.md): In Settings → API keys → Webhooks, enter an https Endpoint URL, pick Events and click Add endpoint. Copy the signing secret (shown once) and use it to verify the PDFBetter-Signature header.
- [Extract data from many PDFs into a table](https://help.pdfbetter.com/ai/extract-pdf-data-table.md): Go to Extractions, choose a preset or define fields, pick documents and click Extract. Review each value against the quote it came from, approve or correct it, then export XLSX, CSV or JSON. Each page read costs 1 credit.
- [Make a scanned PDF editable and searchable with OCR](https://help.pdfbetter.com/editing/make-scanned-pdf-editable-ocr.md): Open the OCR PDF tool, upload the scan, pick the Languages and pages and click Recognize text. The result looks identical but has a searchable text layer; click Open in the editor to change its text. OCR is counted per page recognized.

---
Source: https://help.pdfbetter.com/automation/saved-workflows/ · PDFBetter Help Center · Last checked against the product 2026-10-18. Questions: support@pdfbetter.com