Debugging Workflows
If a workflow is not working, you can debug it in two ways:
- Ask your AI executive assistant (Sulla) to investigate and fix it
- Run manual debugging steps yourself
Fastest path: ask Sulla to debug it
When something breaks, tell Sulla exactly:
- What workflow failed
- What you expected to happen
- What actually happened
- Any error message shown
- Which step/node appears to fail
Sulla can then test and debug the workflow, suggest fixes, and apply revisions.
Prompt template for AI-assisted debugging
Use a message like:
"Debug my workflow called [workflow name]. It should [expected behavior], but instead [actual behavior]. The failure appears at [step/node]. Error message: [error]. Please test and revise the workflow."
Manual workflow debugging steps
If you want to debug manually, use this sequence:
- Open the workflow in the Automation/Workflow system.
- Check the trigger configuration first (schedule, webhook, or event source).
- Run the workflow in test mode.
- Inspect each step output to find the first failing step.
- Validate credentials/integrations used by that step.
- Confirm expected input fields exist and are correctly mapped.
- Re-test after each small change.
Common failure points
- Missing or expired service credentials
- Wrong field mapping between steps
- Trigger misconfiguration (wrong schedule, wrong event, disabled trigger)
- Invalid filters or conditions
- Rate limits or provider-side API errors
Best practice for fixing issues
- Change one thing at a time.
- Re-run with a small test payload.
- Confirm the failing step is fixed before moving on.
- Ask Sulla to optimize the workflow after it is stable.
When to use AI vs manual
- Use AI-first for speed and automatic troubleshooting.
- Use manual debugging when you want deeper control over specific nodes/logic.
Most users should start by asking Sulla to debug, then review the updated workflow and request follow-up revisions if needed.