Skip to main content
Get started with Visual Workflow Builder in minutes. Build your first workflow on the dashboard and execute it via API.

​
Prerequisites

Before you begin, make sure you have:
  • A Dari account at app.usedari.com
  • Your API key from the dashboard
  • The Dari Python SDK installed: pip install dari

​
Step 1: Create Your First Workflow

  1. Log in to app.usedari.com
  2. Navigate to Workflows in the sidebar
  3. Click Create New Workflow
  4. Give your workflow a name (e.g., “My First Workflow”)

​
Step 2: Add Workflow Steps

Add steps to your workflow using the visual builder:

​
Browser Action Step

Click Add Step → Browser Action and enter a natural language instruction: 
Navigate to Google and search for "browser automation"

​
Browser Agent Step

For multi-step tasks, add a Browser Agent: 
Go to example.com, fill out the contact form with my information, and submit it

​
Browser Code Step

For precise data extraction, add a Browser Code step: 
Extract all product prices from the current page

​
Step 3: Define Input Variables

Make your workflow reusable by adding input variables:
  1. Click Add Variable in the workflow builder
  2. Name your variable (e.g., search_query)
  3. Use it in your steps with {{search_query}}
Example step with variable: 
Search Google for "{{search_query}}"

​
Step 4: Save and Get Workflow ID

  1. Click Save in the top right
  2. Click Publish to make it available via API
  3. Copy your Workflow ID from the workflow details

​
Step 5: Execute via API

Now execute your workflow programmatically: 
from dari import Dari

client = Dari(api_key="YOUR_API_KEY")

# Start the workflow
result = client.start_workflow(
    workflow_id="YOUR_WORKFLOW_ID",
    input_variables={
        "search_query": "AI automation tools"
    }
)

print(f"Execution ID: {result['workflow_execution_id']}")
print(f"Status: {result['status']}")

​
Step 6: Monitor Execution

Check the status and get results: 
# Get execution details
execution = client.get_execution_details(
    execution_id=result['workflow_execution_id']
)

print(f"Status: {execution['status']}")
print(f"Output: {execution.get('output', {})}")

​
Your First Complete Workflow

Here’s a complete example that creates a meeting:

​
In the Dashboard:

Create a workflow with these steps:
  1. Step 1 (Browser Agent): 
    Navigate to Google Calendar and create a meeting titled "{{meeting_title}}"
scheduled for {{date}} at {{time}}
  2. Step 2 (Browser Code): 
    Get the meeting URL from the page

​
In Your Code:

from dari import Dari
import time

client = Dari(api_key="YOUR_API_KEY")

# Start workflow
result = client.start_workflow(
    workflow_id="YOUR_WORKFLOW_ID",
    input_variables={
        "meeting_title": "Team Standup",
        "date": "tomorrow",
        "time": "10:00 AM"
    }
)

execution_id = result['workflow_execution_id']
print(f"Workflow started: {execution_id}")

# Wait for completion
while True:
    execution = client.get_execution_details(execution_id=execution_id)
    status = execution['status']

    if status in ['completed', 'failed', 'timeout']:
        break

    time.sleep(5)

# Get results
if status == 'completed':
    meeting_url = execution.get('output', {}).get('meeting_url')
    print(f"Meeting created: {meeting_url}")
else:
    print(f"Workflow {status}")

​
Common Patterns

​
Pattern 1: Data Extraction

# Workflow: Extract product data from e-commerce site
result = client.start_workflow(
    workflow_id="product-scraper",
    input_variables={
        "product_url": "https://example.com/products/123"
    }
)

# Get extracted data
execution = client.get_execution_details(execution_id=result['workflow_execution_id'])
product_data = execution.get('output', {})

​
Pattern 2: Form Submission

# Workflow: Fill and submit forms
result = client.start_workflow(
    workflow_id="form-filler",
    input_variables={
        "name": "John Doe",
        "email": "[email protected]",
        "message": "Hello from Dari!"
    }
)

​
Pattern 3: Multi-Page Navigation

# Workflow: Navigate through multiple pages
result = client.start_workflow(
    workflow_id="multi-page-workflow",
    input_variables={
        "start_url": "https://example.com",
        "search_term": "products",
        "max_pages": 5
    }
)

​
Next Steps

Choosing the Right Step Type

Learn when to use Actions, Agents, or Code

Browser Actions

Quick 2-step browser interactions

Browser Agents

Multi-step autonomous workflows

Browser Code

Custom JavaScript for precise control

​
Troubleshooting

Check that:
  • Your workflow is published (not in draft mode)
  • All required input variables are provided
  • Your API key is valid
  • Increase timeout: timeout_minutes=30
  • Simplify complex workflows into smaller steps
  • Check if the target website is responsive
  • Use Browser Code for reliable element selection
  • Add wait conditions in your workflow steps
  • Verify the page loads completely before interaction
  • Ensure your Browser Code step has a return statement
  • Check that extraction selectors are correct
  • Review execution logs in the dashboard

​
Need Help?

For questions or support, contact us at [email protected].