Understanding Workflows

What Are Workflows?

Workflows in ADM are pre-built, multi-step intelligent processes that help you perform complex data management tasks through an interactive, guided experience. Instead of manually configuring each step, workflows combine AI-powered automation with human validation to streamline tasks such as policy creation, asset classification, and data quality remediation.

Key Characteristics

Workflows represent a shift from one-time queries to orchestrated, multi-stage operations that deliver end-to-end outcomes.

Feature	Description
Template-Based	Begin from pre-configured templates built for specific use cases.
Interactive Execution	Combine automated processing with user checkpoints and approvals.
AI-Assisted	Intelligent agents analyze context and make data-driven recommendations.
Guided Experience	Clear, step-by-step progression with visible status indicators.
Configurable	Customize workflow behavior using prompts, parameters, and optional scheduling.

Workflow vs. Simple Query

Aspect	Simple Query	Workflow
Complexity	Single-step request	Multi-step coordinated process
User Interaction	One-time	Interactive with validation points
Duration	Seconds	Minutes to hours
Validation	Automatic	User-reviewed at key stages
Output	Response or insight	Configuration, policy, or applied action
Resumability	Not applicable	Can be paused and resumed

Accessing Workflows

Workflow Library

To access workflows:

Navigate to Workflows in ADM.
The Workflow Library opens, displaying all available templates.

The library provides:

Browse Templates – View available workflow templates with descriptions.
Search Functionality – Find workflows by name or use case.
Template Details – Each template displays:
- Workflow name and icon
- Short description
- Number of steps
- Category or tag
- Use This Template button

Available Workflow Templates

Template Name	Description
Intelligent Asset Classification and Compliance Workflow	Classifies data assets across taxonomies with confidence scoring and generates detailed reports.
Business Context Generation Workflow	Analyzes metadata to produce business context and documentation for assets.
Data Quality Remediation Workflow	Identifies data quality issues, determines root causes, and applies automated fixes.
Policy Creation Workflow	Analyzes policy requirements and automatically generates data quality policies.
Policy Description Workflow	Summarizes and documents existing policies for reporting or audit readiness.

Creating and Executing a Workflow

Starting a Workflow

In the Workflow Library, click Use This Template.
The Create New Workflow dialog appears.
Provide the following details:
- Workflow Name – Enter a descriptive name for the instance.
- Description (optional) – Add context or purpose.
- Enable Scheduler – Toggle ON to schedule automatic runs.
Click Save Workflow.

Your new workflow opens in the Execution View.

Workflow Step Types

Step Type	Purpose
User Request	Capture user instructions or parameters.
Intent Validation	Confirm requirements and identify required resources.
Processing	Execute automated or AI-driven analysis.
Generation	Produce configurations, reports, or artifacts.
Review and Save	Present results for approval before saving or applying.

Managing Workflow Runs

Workflow Dashboard

The Workflows page serves as the central management dashboard.

Column	Description
Instance Name	Custom name for the workflow (clickable for details).
Last Run	Timestamp of the most recent execution.
Last Run Status	Indicates Success, Failed, or In Progress.
Last Updated	Date of the last modification.
Created	Original creation timestamp.
Actions	Edit or delete workflow.

Dashboard Features

Search Bar – Find workflows by name.
Refresh Button – Reloads latest run statuses.
Browse Templates – Opens the Workflow Library.
Pagination and Row Settings – Adjust view preferences.

Viewing Execution Details

Click any workflow name to view:

Execution Timeline – Visual flow of completed steps.
Step Details – Inputs, logs, outputs, and execution time.
Error Messages – Visible for failed steps.
Run History – Records of all previous executions.

Workflow States

State	Symbol	Description
Not Started		Step hasn't begun execution Prerequisites not yet met Waiting for previous step to complete
Running		Step is actively executing AI agents processing data Background operations in progress May take several seconds to minutes depending on complexity
Waiting for Input		Step requires user review or decision Validation checkpoint reached User must provide input to continue
Completed		Step finished successfully Output generated and available Ready to proceed to next step
Failed	X	Step encountered an error Execution stopped Requires user intervention May offer retry or skip options

Working with Workflow Results

Reviewing Outputs

Outputs vary by workflow type and may include:

Generated data quality policies
Detailed classification reports
Automated remediation plans
Enriched business context and metadata

Saving and Applying Results

Review outputs after execution.
Modify configurations if needed.
Validate correctness.
Click Save to persist results.
Verify that changes were applied successfully in ADM.

Best Practices

Stage	Recommendations
Before Starting	Define objectives, identify data assets, confirm permissions, estimate duration.
During Execution	Monitor progress, review checkpoints, provide clear input, and avoid navigation during critical steps.
After Completion	Verify results, document changes, save reusable workflows, review logs for any issues.

Troubleshooting Failed Workflows

Check Error Message – Identify the failing step.
Verify Prerequisites – Ensure data sources and permissions are valid.
Retry Step – Many failures are transient.
Adjust Input – Refine parameters if validation fails.
Contact Support – Provide workflow name, step logs, and context for assistance.

Workflow Scheduler

The Scheduler enables automated execution of workflows at predefined intervals.

Action	Description
Enable During Creation	Toggle Enable Scheduler when setting up a workflow.
Automatic Runs	Executes on schedule without manual input.
Notebook Integration	Can trigger scheduled notebook runs automatically.
Disable Option	Turn off to stop automatic execution.

Note Scheduler is disabled by default. A message appears: “Scheduler is disabled. This notebook will not run automatically.”

Workflow Templates by Use Case

Template Type	Purpose
Asset Classification	Validates and classifies assets with confidence scoring and report generation.
Policy Creation	Analyzes requirements, selects assets, and creates rules for data quality policies.
Data Quality	Detects quality issues, analyzes error patterns, and suggests automated remediation.
Business Context	Collects metadata, execution history, and samples to generate descriptive documentation.

Workflow Execution Tips

Prompt Customization

Be specific and concise when defining objectives.
Reference relevant datasets or rules.
Indicate constraints, preferences, or success conditions.
Use natural language — ADM understands context intuitively.

Managing Long-Running Workflows

Workflows with multiple steps may take several minutes.
You can navigate away and return later; ADM preserves progress.
Real-time status indicators show running or failed steps.

Workflow Versioning

Each execution is stored as a unique run.
Compare outputs across versions to refine performance.
Use successful runs as templates for future automation.

Last updated on

Was this page helpful?