Skip to main content
Workflows let you automatically extract structured data from documents using AI. Define what data you need, upload documents, and get structured results.

Workflows Overview

The Workflows page has two tabs:
  1. Workflow Runs - See all documents that have been processed
  2. Extraction Templates - Create and manage extraction templates

Workflow Runs Tab

This tab shows all documents submitted for data extraction.

Table Columns

ColumnDescription
FilenameName of the uploaded document
Extraction TemplateWhich template was used for extraction
StatusPending, Processing, Completed, or Failed
CreatedWhen the document was submitted to EmberQA

Filtering Workflow Runs

Click the Filters section to expand it. Available filters:
FilterDescription
StatusFilter by Pending, Processing, Completed, or Failed
Extraction TemplateShow only runs using a specific template
FilenameSearch by document filename
After setting filters, click Apply Filters. Use Clear All to reset.

Viewing Run Details

Click any row to open the detail popup. The popup shows: Basic Information
  • Filename
  • Extraction Template used
  • Created timestamp
Failure Reason (if applicable)
  • If the extraction failed, you’ll see the error message explaining why
Metadata
  • Any custom metadata attached to the document (shown as key-value pairs)
Extracted Data
  • A table showing each property defined in the template
  • For each property: the extracted value and the AI’s reasoning for that extraction
  • If extraction is still in progress, you’ll see a “Data extraction is in progress…” message
Supporting Documents
  • If the workflow included supporting documents, you can view them here
Use the Previous/Next buttons to navigate between workflow runs.

Extraction Templates Tab

Extraction templates define what data to pull from your documents. Think of them as forms that the AI fills out by reading your documents.

Template List

The list shows all your extraction templates with:
ColumnDescription
Template NameDisplay name for the template
IDUnique identifier (used in API calls)
PropertiesChips showing each field the template extracts
ActionsView or Edit buttons

Creating a New Template

  1. Click Create New Template
  2. Enter a Template Name
  3. Add Properties - these are the fields you want to extract
For each property, you’ll define:
  • Property Name - The field name (e.g., “invoice_number”, “customer_name”)
  • Description - Instructions telling the AI what to look for
  • Type - Either:
    • Extract Text - AI extracts free-form text
    • Select One - AI picks from predefined options you provide
  1. Click Save when done

Property Types Explained

Extract Text Use this when you want the AI to find and extract text from the document. Examples:
  • Customer name
  • Invoice number
  • Contract date
  • Total amount
Select One Use this when the answer should be one of a fixed set of options. For each option, provide:
  • Option Key - The value that will be returned (e.g., “approved”)
  • Option Value - Description of when to select this option (e.g., “Document shows approval signature”)
Examples:
  • Status: Approved, Denied, Pending
  • Document Type: Invoice, Quote, Receipt
  • Priority: High, Medium, Low

Editing Templates

  1. Find the template in the list
  2. Click Edit
  3. Make your changes
  4. Click Save

Viewing Template Details

Click View on any template to see its full configuration without entering edit mode.

Submitting Documents for Extraction

Documents can be submitted via:
  • Bulk Upload page - Manual upload through the UI
  • API - Programmatic submission (see Workflow API)
When submitting, you’ll select which Extraction Template to use. The AI will then process the document and extract data according to that template’s properties.