Workflows

Workflows capture a business bottleneck, the systems in scope, and the steps Trame will take to resolve it. You can run them on demand or trigger them automatically.

Building a workflow

AI-Powered Creation

Go to Workflows → Create Workflow.
Describe your automation goal in plain language using the intelligent workflow builder.
Trame’s AI generates a complete workflow specification including:
- Clear title and detailed description
- Required connectors and integrations
- Trigger suggestions based on your description
- Configuration fields and extracted values
- Feasibility assessment and recommendations
Review and customize the generated workflow before saving.
Fill in optional configuration details using AI research assistance.
Save the workflow. It starts in Draft status for safe testing.

Workflow Generation Process

The AI creation process includes:

Natural Language Analysis: AI understands your automation goal
System Integration: Identifies required connectors and tools
Trigger Intelligence: Suggests appropriate workflow triggers
Configuration Extraction: Pulls specific values from your description
Detail Requests: Identifies additional information for optimal execution
Real-time Preview: See the workflow taking shape as it’s generated

Status and approvals

Draft: best for iteration; runs pause for approval before executing.
Pilot: allows triggers, but still pauses for approval so you can supervise.
Live: executes immediately when triggered or started manually.
Paused: ignores triggers until resumed.

Triggers and webhooks

Triggers come from connected toolkits (e.g., a Slack event, CRM update, or email).
Choose a trigger when setting up the workflow; only connected toolkits can supply triggers.
Paused/Draft workflows ignore trigger events; Pilot/Live workflows start as soon as the trigger fires (Pilot will still wait for approval).
You can always run a workflow manually from its detail page for tests or urgent needs.

Execution tracking and intelligence

Real-time Monitoring

Each workflow execution includes comprehensive tracking:

Step-by-step Progress: Real-time updates as the workflow executes
Tool Execution Timeline: Record of which integrations were used and when
Duration Tracking: Live timing from start to completion
Status Updates: Current execution state with detailed progress information

Execution Data

Each run stores detailed information:

Planned Steps: The AI-generated execution plan before starting
Actual Execution: Step-by-step log of what actually happened
Tools Used: Complete record of integrations and actions taken
Approval History: Who approved runs and when decisions were made
Input Data: Trigger payload or manual run parameters
Output Results: Final results and any generated content

Intelligent Learning

Failed executions trigger automatic analysis:

Failure Type Classification: Objective failures vs technical errors
Root Cause Analysis: AI determines why the workflow couldn’t complete
Improvement Suggestions: Specific, actionable recommendations for fixes
Missing Capabilities: Identification of required tools or permissions
Configuration Recommendations: Suggested workflow setting improvements

Success Pattern Recognition

Successful executions contribute to learning:

Discovered Context: Useful IDs, URLs, and resources for future runs
Successful Tool Combinations: Effective integration patterns
Timing Optimization: Optimal execution timing and conditions
Pattern Sharing: Successful approaches applied to similar workflows

Webhook integration

Custom Webhook Endpoints

Each workflow automatically gets a unique webhook URL:

Unique URLs: Cryptographically secure endpoints like /api/hooks/invoice-processor-k3m9
Automatic Generation: Webhook URLs created automatically when workflows are saved
Organization Scoped: Webhooks are isolated to your organization
HTTP POST Support: External systems trigger workflows via HTTP POST requests

Webhook Security

Multiple security layers protect webhook endpoints:

Optional Secrets: Configure shared secrets for HMAC signature verification
Unique Identifiers: Each workflow gets a unique, unpredictable webhook slug
Request Validation: Incoming payloads validated before workflow execution
HTTPS Only: All webhook requests must use secure HTTPS protocol

Payload Processing

Webhook requests become workflow input data:

JSON Payloads: Structured data accessible within workflow execution
Header Information: Request headers available for conditional logic
Metadata Access: Timestamp, source IP, and request metadata
Data Validation: Optional schema validation for payload structure

Integration Examples

Common webhook use cases:

E-commerce: Order notifications from Shopify, WooCommerce
CRM Systems: Lead updates from Salesforce, HubSpot
Support Systems: Urgent ticket notifications from Zendesk
Payment Processing: Transaction events from Stripe, PayPal
Custom Applications: Events from internal business systems

Workflow configuration best practices

Keep sensitive data in workflow configuration fields rather than trigger descriptions
Use specific connector permissions to limit workflow access scope
Test workflows thoroughly in Draft mode before promoting to Live
Monitor execution logs and apply AI-suggested improvements regularly
Document workflow purpose and configuration for team collaboration