Skip to main content

Create a New Workflow

This guide walks you through creating your first workflow in Vectense Platform, from initial setup to testing and deployment.

Getting Started​

Prerequisites​

Before creating a workflow, ensure you have:

  • Active Workspace: Access to a workspace with workflow creation permissions
  • AI Model: At least one configured language model
  • Basic Understanding: Familiarity with your automation requirements

Access Workflow Creation​

  1. Navigate to Workflows: Click "Workflows" in the main navigation
  2. Create New: Click "Create a new workflow" button
  3. Choose Creation Method: Select how you want to build your workflow

Creation Methods​

Step 1: Choose AI Assistant

  • Select "Assistant" from the creation options
  • This provides guided workflow creation with AI help

Step 2: Select a Model

  • Choose which AI model the assistant should use
  • The model will help design your workflow structure
  • Ensure you have at least one model configured

Step 3: Describe Your Process Answer the assistant's questions about your business process:

What does your AI-Process do?

  • Provide a clear description of your workflow's purpose
  • Examples: "Process customer support emails", "Analyze sales reports", "Extract invoice data"

What starts your AI-Process?

  • Describe the trigger event
  • Examples: "When a new email arrives", "Every Monday at 9 AM", "When a file is uploaded"

What should your AI-Process do?

  • Describe the step-by-step process
  • Be specific about inputs, processing, and outputs
  • Include any decision points or conditional logic

What should be the result?

  • Describe the final outcome
  • Examples: "Send response email", "Update database", "Create summary report"

Step 4: Review Generated Workflow

  • The AI assistant creates a workflow based on your answers
  • Review the proposed triggers, steps, and actions
  • Make adjustments as needed

Step 5: Customize and Deploy

  • Fine-tune the generated workflow
  • Test with sample data
  • Activate when ready

Method 2: Manual Creation​

Step 1: Choose Manual Creation

  • Select "Start yourself" for full control over workflow design
  • Ideal for experienced users or complex requirements

Step 2: Basic Information Configure workflow basics:

  • Name: Descriptive workflow name
  • Description: Optional detailed description
  • Active Status: Start as inactive for testing

Step 3: Add Triggers Configure what starts your workflow:

Scheduled Trigger

  • Select "Scheduled Task" trigger
  • Set cron expression for timing
  • Configure timezone if needed
  • Set output variable name for timestamp

Webhook Trigger

  • Select "HTTP-Webhook Listener"
  • Configure shared key for security
  • Set memory variables for headers and body
  • Note the webhook URL for external systems

File System Trigger

  • Select "File System Monitor"
  • Configure directory path to watch
  • Set file pattern filters
  • Define action on file changes

Step 4: Add Processing Steps Add steps to process data and make decisions:

LLM Analysis Step

  • Add "LLM Prompt" action
  • Configure persona (role, goals, skills, communication style)
  • Select AI model
  • Write system instruction
  • Set input context from previous steps
  • Define output variables

HTTP Request Step

  • Add "HTTP Request Runner" action
  • Configure URL and HTTP method
  • Set request body from memory variables
  • Define result storage variable

File Operation Step

  • Add "File Write" action
  • Configure file path and format
  • Set content from memory variables
  • Define success/failure handling

Step 5: Configure Workflow Logic Set up data flow and conditions:

Memory Variables

  • Define input variables from triggers
  • Set intermediate processing variables
  • Configure output variables for actions

Conditional Logic

  • Add conditions to control workflow flow
  • Set up error handling paths
  • Configure retry logic for failed steps

Step 6: Test and Deploy

  • Use manual testing to verify workflow
  • Debug any configuration issues
  • Activate workflow when ready

Method 3: Marketplace Templates​

Step 1: Browse Templates

  • Select "Marketplace Templates" option
  • Browse available workflow templates
  • Filter by category or use case

Step 2: Select Template

  • Choose a template that matches your needs
  • Review template description and requirements
  • Check required integrations and models

Step 3: Customize Template

  • Import template to your workspace
  • Modify triggers, steps, and configurations
  • Adapt to your specific requirements
  • Update model and integration settings

Workflow Configuration Details​

Persona Configuration​

When using AI steps, configure the persona:

Role

  • Define what role the AI should take
  • Examples: "Customer Service Representative", "Data Analyst", "Content Reviewer"

Goals

  • Specify what the AI should accomplish
  • Examples: "Respond helpfully to customer questions", "Extract key data points accurately"

Skills

  • List specific capabilities
  • Examples: "Email writing", "Document analysis", "Data validation"

Communication Style

  • Define response tone and format
  • Examples: "Professional and concise", "Friendly and conversational", "Technical and detailed"

Guardrails

  • Set boundaries and limitations
  • Examples: "Never share personal information", "Always maintain professional tone", "Stick to factual information"

Memory Management​

Configure data flow between steps:

Input Variables

  • Variables that receive data from triggers
  • Examples: emailContent, fileName, scheduledTime

Processing Variables

  • Intermediate data storage
  • Examples: analysisResult, extractedData, validationStatus

Output Variables

  • Final results for external use
  • Examples: responseMessage, processedFile, summaryReport

Integration Settings​

Configure connections to external systems:

API Credentials

  • Store API keys securely
  • Use secured field types for sensitive data
  • Test connections before deployment

File System Access

  • Configure directory permissions
  • Set file pattern filters
  • Define access credentials if needed

Database Connections

  • Configure connection strings
  • Set query parameters
  • Define data mapping

Testing Your Workflow​

Manual Testing​

Before activating automatic triggers:

Test with Sample Data

  • Use "Manual Run" feature
  • Provide test input data
  • Verify each step executes correctly
  • Check output quality and format

Debug Configuration Issues

  • Review step-by-step execution logs
  • Check memory variable values
  • Verify integration connections
  • Validate AI model responses

Performance Testing

  • Test with different data sizes
  • Measure execution times
  • Check resource usage
  • Verify error handling

Gradual Deployment​

Deploy workflows incrementally:

Start Inactive

  • Create workflow in inactive state
  • Test thoroughly before activation
  • Make adjustments based on test results

Limited Scope Testing

  • Start with restricted triggers
  • Monitor initial executions closely
  • Expand scope after validation

Full Deployment

  • Activate all triggers
  • Monitor performance metrics
  • Set up alerting for failures

Monitoring and Optimization​

Execution Monitoring​

Track workflow performance:

  • Job Status: Monitor active and completed jobs
  • Success Rates: Track execution success/failure ratios
  • Performance Metrics: Measure execution times and resource usage
  • Error Analysis: Review failed executions and error patterns

Cost Management​

Monitor and optimize costs:

  • AI Model Usage: Track token consumption and costs
  • Processing Time: Monitor compute resource usage
  • Storage Usage: Track data storage requirements
  • Optimization Opportunities: Identify cost reduction possibilities

Workflow Maintenance​

Keep workflows running optimally:

  • Regular Testing: Periodic testing with various input scenarios
  • Performance Monitoring: Track metrics over time
  • Configuration Updates: Update integrations and models as needed
  • Documentation: Maintain current workflow documentation

Common Workflow Patterns​

Document Processing​

  1. Trigger: File upload or email attachment
  2. AI Analysis: Extract key information using LLM
  3. Data Validation: Verify extracted data completeness
  4. Action: Store in database or send to external system

Customer Service Automation​

  1. Trigger: Incoming support ticket or email
  2. Classification: Categorize request using AI
  3. Knowledge Retrieval: Find relevant information from knowledge base
  4. Response Generation: Create appropriate response
  5. Action: Send response or route to human agent

Data Integration​

  1. Trigger: Scheduled execution or data change
  2. Data Retrieval: Fetch data from source systems
  3. Transformation: Process and validate data
  4. Quality Check: Verify data integrity
  5. Action: Load data into target system

Troubleshooting​

Common Issues​

Workflow Won't Start

  • Check trigger configuration
  • Verify user permissions
  • Ensure workflow is active
  • Review workspace license status

Steps Failing

  • Verify integration configurations
  • Check API keys and credentials
  • Review input data format
  • Test external system connectivity

Unexpected Results

  • Review AI model prompts and persona
  • Check memory variable data flow
  • Verify conditional logic
  • Test with different input data

Getting Help​

  • Documentation: Reference specific integration guides
  • Community: Ask questions in user forums
  • Support: Contact technical support for complex issues
  • Logs: Review detailed execution logs for debugging

Next Steps​

After creating your workflow:

  1. Edit and Refine: Learn to modify and optimize workflows
  2. Monitor Execution: Track performance and results
  3. Add Integrations: Connect to additional systems
  4. Create Additional Workflows: Build more automations for your organization

Your workflow is now ready to automate your business processes intelligently!