Overview
Connect Data Hubs to your workflows to provide contextual knowledge from your organization's curated resources. This integration enables your workflows to reference internal documentation, policies, and data alongside web scraping and other workflow steps.
Learning Objectives:
Add Data Hub steps to new or existing workflows
Configure collection selection and extraction queries
Apply best practices for relevant context selection
Troubleshoot common Data Hub integration issues
Prerequisites
You'll Need:
Access to a workspace with workflow builder enabled
At least one Data Hub with collections containing relevant documents
Edit permissions for the workflow you're modifying
Access Workflow Builder
Insert Data Hub Integration
Open Workflow Editor
Navigate to Workflows section
In left panel, click "Workflows"
Workflow center interface loads
Open or create workflow
New workflow: Click "Create Workflow" (top-right)
Existing workflow: Click workflow name from list
Access step insertion
In workflow canvas, click plus (+) icon where you want to add step
"Add Step" menu appears with available step types
Result: Workflow builder is open and ready for step configuration.
Configure Step Types
Chat Step (AI Generation & Analysis)
Select Chat Step
Click "Chat" in Add Step menu
Configure AI settings
Select model: Choose GPT-4, Gemini, or other available model
Write prompt: Enter instruction in prompt field
Use variables: Reference previous step outputs with
{{step_name.output}}
Set conversation style: Adjust the style (Creative, Balanced, Precise)
Click "Save Workflow"
Go to the "Run" tab and click on the Run workflow button
Check the "Results" section to review the output of every step
Example Prompt:
Analyze students grades: {{upload_grades_csv}}
Identify top 3 studentsUse variables: Reference previous step outputs with
{{step_name.output}}Tip: To find variable ID, click the source step in canvas → Copy the variable ID displayed below the step name
Result: AI analyzes input and generates content based on your prompt.
Best For: Content generation, data analysis, decision-making, text transformation
Best Practices
Start simple - Begin with 2-3 steps; add complexity gradually
Use variables - Make workflows dynamic and reusable
Chain steps logically - Each step should build on previous outputs
Test thoroughly - Run workflow after each change to verify output
Monitor performance - Check execution times; optimize slow steps
Handle errors - Add fallback steps for critical workflows
Document your workflow - Add descriptions so team members understand the flow
Common Questions
Q: What happens if Data Hub is empty?
A: Workflow step completes but returns no context. Add documents to Data Hub or adjust query.
Q: Can I use multiple Data Hubs in one workflow?
A: Yes—add multiple Datastore steps or select multiple collections in single step.
Q: Do Data Hub updates affect running workflows?
A: Yes—workflows automatically use current Data Hub content on each run.
Q: Why are my Data Hub results irrelevant?
A: Query too broad or wrong collection selected. Refine query specificity and verify collection relevance.
Q: How many steps can a workflow have?
A: No hard limit, but workflows with 10+ steps can become difficult to debug. Consider splitting complex processes.
Q: Can I reorder steps after adding them?
A: Yes—drag and drop steps in the canvas to reorder. Variables automatically update.
Q: Can I test individual steps?
A: No—test full workflow only. Isolate steps in separate test workflows for debugging.