Skip to main content

​
Overview

Once you’ve built a workflow, it’s time to run it. This guide covers everything from manual testing to monitoring production executions.

​
Running a workflow

​
Manual execution

The quickest way to test your workflow:
  1. Click the App button in the top-right corner
  2. A panel opens showing your workflow’s input form
  3. Fill in any required inputs
  4. Click Generate

​
Execution modes

ModeDescriptionUse case
Run onceExecute a single timeTesting, one-off tasks
Run with inputsProvide custom inputsProcessing specific data
ScheduledRun automaticallyRegular reports, monitoring

​
Monitoring execution

​
Real-time progress

While your workflow runs, you can see:
  • Current node: Which node is executing
  • Progress: How many nodes completed
  • Duration: Time elapsed
  • Status: Running, completed, or failed

​
Node status indicators

ColorStatus
🔵 BlueCurrently executing
🟢 GreenCompleted successfully
🔴 RedFailed with error
⚪ GreyNot yet executed

​
Viewing outputs

Click on any completed node to see:
  • Output data: The actual result
  • Execution time: How long it took
  • Logs: Detailed execution logs

​
Debugging workflows

​
Common issues

Cause: The previous node didn’t return expected data.Solution:
  1. Check the previous node’s output
  2. Verify your variable path is correct
  3. Add a Conditional node to handle missing data
Cause: External service issue or invalid configuration.Solution:
  1. Check your API credentials
  2. Verify the endpoint URL
  3. Test the API directly with curl or Postman
  4. Check rate limits
Cause: Too many items or infinite loop condition.Solution:
  1. Add a maximum iteration limit
  2. Check your loop exit condition
  3. Process data in batches
Cause: Processing takes too long.Solution:
  1. Increase timeout in workflow settings
  2. Optimize slow nodes
  3. Run heavy operations in parallel

​
Using run history

Access the complete history of all workflow executions:
  1. Open your workflow
  2. Click on the Run History tab
  3. View all past executions with their status, duration, and outputs
Run History tab showing past workflow executions
The run history table displays:
  • ID: Unique identifier for each run
  • Created At: Date and time of execution
  • Status: Done, Failed, or Running
  • Duration: Total execution time
  • Workflow Version: Draft or Published version used
  • Steps: Click the eye icon to view detailed step information
  • Main Output: Quick access to the final result

​
Viewing step details

Click the Steps icon (👁) on any run to see a detailed breakdown of each node’s execution:
Workflow steps panel showing detailed execution
The step details panel shows:
  • Workflow steps: List of all executed nodes with status indicators
  • Execution time: Duration for each individual node
  • Compact mode: Toggle to show/hide details
  • Render/Code view: Switch between formatted output and raw data
  • Copy/Download: Export the output data
Use Compact mode to quickly scan through all steps, then click on a specific step to expand its details.

​
Step-by-step debugging

For complex issues:
  1. Isolate the problem: Disable nodes after the failing one
  2. Check inputs: Verify the failing node receives correct data
  3. Test manually: Run the operation outside Draft & Goal
  4. Review connections: Ensure data flows correctly
  5. Compare runs: Use history to compare successful vs failed executions

​
Debugging AI Agents

AI Agent nodes require special debugging because they make autonomous decisions. Draft & Goal provides a dedicated Agent Trace feature to understand exactly what your agent did.

​
Accessing Agent Traces

When viewing workflow steps, AI Agent nodes display a clock icon (🕐) next to their execution time. Click this icon to open the trace viewer:
Agent trace button in workflow steps

​
Understanding Agent Traces

The Agent Trace panel shows the complete conversation between the AI and its tools:
Agent trace detail showing conversation history
The trace includes:
Message TypeDescription
SystemThe initial instructions given to the agent
HumanThe task or prompt your workflow sent
AIThe agent’s reasoning and tool calls
ToolResults returned by tools the agent used
Switch between Conversation view for a readable format and Raw JSON for the complete technical data.

​
Common agent debugging scenarios

Check: Review the System message - are instructions clear?Solution: Refine your agent prompt to be more specific about when to use each tool.
Check: Look for repeated tool calls with same parameters.Solution: Add clearer success criteria or a maximum iteration limit in your prompt.
Check: Examine the Tool responses - did they return expected data?Solution: Verify tool configurations and test tools independently.

​
Error handling

​
Built-in error handling

Configure how your node handles failures: Workflow settings → Error Handling
OptionBehavior
Stop on errorHalt workflow immediately
Continue on errorSkip failed node, continue
RetryAttempt failed node again

​
Using the Fail node

Gracefully handle expected errors:
  1. Add a Conditional node to check for problems
  2. Route to a Fail node with a clear message
  3. The workflow stops with your custom error
If {{response.status}} != 200
  → Fail: "API returned error: {{response.error}}"

​
Performance optimization

​
Identify bottlenecks

After running, check execution times:
  1. Look at the duration for each node
  2. Find the slowest nodes
  3. Optimize or parallelize them

​
Parallel execution

Run independent operations simultaneously: 
        ┌→ Scrape Site A →┐
Input →│                   │→ Merge → Output
        └→ Scrape Site B →┘
Instead of: 
Input → Scrape Site A → Scrape Site B → Output

​
Caching strategies

Reduce redundant API calls:
  • Use Google Sheets as a cache layer
  • Check if data exists before fetching
  • Set appropriate cache TTLs

​
Batch processing

Process large datasets efficiently:
  1. Use the Loop node with batch settings
  2. Set reasonable batch sizes (10-50 items)
  3. Add delays between batches for rate limits

​
Execution history

​
Viewing past runs

Access the execution history:
  1. Go to Workflows in the main menu
  2. Click on your workflow
  3. Select the Runs tab
See:
  • Run date and time
  • Duration
  • Status (success/failure)
  • Input values used

​
Re-running workflows

To repeat a previous execution:
  1. Find the run in history
  2. Click Run again
  3. Optionally modify inputs
  4. Execute

​
Exporting results

Download execution data:
  1. Click on a completed run
  2. Select Export
  3. Choose format (JSON, CSV)

​
Scheduled execution

​
Setting up schedules

Run workflows automatically:
  1. Open workflow settings
  2. Go to Schedule
  3. Choose frequency:
ScheduleExample
HourlyEvery hour at :00
DailyEvery day at 9:00 AM
WeeklyEvery Monday at 8:00 AM
CustomCron expression

​
Cron expressions

For advanced scheduling: 
┌───────────── minute (0-59)
│ ┌───────────── hour (0-23)
│ │ ┌───────────── day of month (1-31)
│ │ │ ┌───────────── month (1-12)
│ │ │ │ ┌───────────── day of week (0-6)
│ │ │ │ │
* * * * *
Examples:
  • 0 9 * * 1 - Every Monday at 9:00 AM
  • 0 */6 * * * - Every 6 hours
  • 30 8 1 * * - 1st of each month at 8:30 AM

​
Monitoring scheduled runs

Check scheduled execution status:
  1. View upcoming runs in the Schedule tab
  2. Get notifications on failures
  3. Review execution history

​
Next steps

Teams

Collaborate with your team

API Reference

Trigger workflows via API