Skip to main content
Elementum’s AI-powered automation actions enable your workflows to understand, analyze, and act on data intelligently. These actions process unstructured information, make intelligent decisions, and transform data automatically within your automation workflows.

​
AI Actions Overview

AI actions receive data through variables and produce structured outputs that can be used by subsequent actions in your automation sequence. Each AI action is designed to handle specific types of data processing and analysis tasks.

AI Classification

Automatically categorize and classify data using AI analysis

AI Summarization

Generate concise summaries of lengthy content and documents

Transform Data with AI

Clean, normalize, and enhance data automatically using AI

AI File Analysis

Extract structured information from documents intelligently

​
AI Classification

Automatically categorize and classify data using AI analysis of content and context.

​
How It Works

AI Classification analyzes content and assigns appropriate categories, tags, or classifications based on patterns and context. The AI understands the meaning and intent of the content to make accurate categorization decisions.

​
Use Cases

Automatically classify incoming support tickets by type, priority, and department.Example Flow:
Email Received → AI Classification → Route to Appropriate Team
Input: Email content from customer supportOutput: Category (Bug, Feature Request, General Support), Priority (High, Medium, Low)
Analyze lead information to determine quality and sales potential.Example Flow:
Lead Form Submitted → AI Classification → Update Lead Score → Route to Sales Team
Input: Lead form data and company informationOutput: Lead Quality (Hot, Warm, Cold), Industry Category, Company Size
Organize documents by type, department, or business process.Example Flow:
Document Uploaded → AI Classification → File in Appropriate Folder → Notify Stakeholders
Input: Document content and metadataOutput: Document Type (Contract, Invoice, Report), Department, Process Category
Analyze customer feedback, reviews, or communications for sentiment and topics.Example Flow:
Customer Feedback Received → AI Classification → Route Based on Sentiment → Create Follow-up Actions
Input: Customer communication or feedback textOutput: Sentiment (Positive, Negative, Neutral), Topic Categories, Urgency Level

​
Variables and Integration

AI Classification outputs structured data that can be used in subsequent automation actions: 
ai_classification.category = "Bug Report"
ai_classification.confidence = 0.92
ai_classification.priority = "High"
ai_classification.department = "Engineering"

​
AI Summarization

Generate concise summaries of lengthy content, extracting key points and insights for quick understanding.

​
How It Works

AI Summarization analyzes long-form content and extracts the most important information into digestible summaries. The AI understands context and relationships to create meaningful condensed versions.

​
Use Cases

Convert detailed reports into executive-level summaries for leadership review.Example Flow:
Monthly Report Generated → AI Summarization → Send to Executive Team → Post to Dashboard
Input: Detailed performance reports, analytics dataOutput: Key metrics, trends, recommendations, action items
Summarize customer feedback to identify common themes and issues.Example Flow:
Customer Survey Responses → AI Summarization → Identify Key Themes → Create Action Plan
Input: Survey responses, feedback forms, support ticketsOutput: Common themes, sentiment summary, priority issues
Extract key decisions and action items from meeting transcripts or notes.Example Flow:
Meeting Notes Uploaded → AI Summarization → Extract Action Items → Create Follow-up Tasks
Input: Meeting transcripts, discussion notesOutput: Key decisions, action items, assigned responsibilities
Create digestible summaries of lengthy documents, articles, or communications.Example Flow:
Research Document Uploaded → AI Summarization → Create Brief → Distribute to Team
Input: Research papers, lengthy reports, technical documentationOutput: Key findings, methodology summary, conclusions

​
Variables and Integration

AI Summarization provides structured summary content: 
ai_summarization.summary = "Key project milestones achieved with 15% budget savings"
ai_summarization.key_points = ["Budget savings achieved", "Timeline on track", "Resource optimization"]
ai_summarization.sentiment = "Positive"

​
Transform Data with AI

Clean, normalize, and enhance data automatically using AI-powered transformation capabilities.

​
How It Works

Transform Data with AI analyzes data patterns and applies intelligent transformations to standardize, clean, and enhance data quality. The AI understands data context and applies appropriate normalization rules.

​
Use Cases

Normalize addresses into consistent formats for better data quality.Example Flow:
Customer Data Imported → Transform Data with AI → Update Customer Records → Validate Addresses
Input: “123 main st, apt 4, NYC, ny 10001”Output: “123 Main Street, Apartment 4, New York, NY 10001”
Standardize phone numbers across different input formats.Example Flow:
Contact Information Updated → Transform Data with AI → Update Contact Records → Verify Numbers
Input: “5551234567”, “(555) 123-4567”, “555.123.4567”Output: “+1 (555) 123-4567”
Normalize company names to prevent duplicate entries.Example Flow:
Lead Data Imported → Transform Data with AI → Check for Duplicates → Merge or Create Records
Input: “Microsoft Corp”, “Microsoft Corporation”, “MSFT”Output: “Microsoft Corporation”
Enhance existing data with additional context and standardization.Example Flow:
Customer Database Updated → Transform Data with AI → Enrich with Industry Data → Update Profiles
Input: Raw customer data with inconsistent formattingOutput: Enriched, standardized customer profiles with industry classifications

​
Variables and Integration

Transform Data with AI provides cleaned and enhanced data: 
transform_data.standardized_address = "123 Main Street, Suite 200, New York, NY 10001"
transform_data.formatted_phone = "+1 (555) 123-4567"
transform_data.company_name = "Microsoft Corporation"
transform_data.confidence_score = 0.95

​
AI File Analysis

Extract structured information from documents intelligently, understanding context and relationships within the content.

​
How It Works

AI File Analysis processes various document types and extracts relevant structured data based on document context. The AI understands different document formats and can identify key information automatically.

​
Use Cases

Extract key terms, parties, dates, and obligations from legal contracts.Example Flow:
Contract PDF Uploaded → AI File Analysis → Create Contract Record → Set Renewal Reminders
Input: Legal contract documentOutput: Parties, effective dates, terms, obligations, renewal dates
Extract vendor information, amounts, line items, and due dates from invoices.Example Flow:
Invoice Uploaded → AI File Analysis → Create Invoice Record → Route for Approval
Input: Invoice PDF or imageOutput: Vendor name, invoice amount, line items, due date, tax amount
Extract skills, experience, and education from candidate resumes.Example Flow:
Resume Uploaded → AI File Analysis → Update Candidate Profile → Score Against Requirements
Input: Resume document in various formatsOutput: Skills, experience, education, contact information, certifications
Extract key metrics, trends, and insights from business reports.Example Flow:
Report Uploaded → AI File Analysis → Extract Key Metrics → Update Dashboard
Input: Business reports, financial statements, analytics reportsOutput: Key metrics, trends, performance indicators, insights

​
Variables and Integration

AI File Analysis provides structured document data: 
ai_file_analysis.contract_parties = ["Company A", "Company B"]
ai_file_analysis.effective_date = "2024-01-15"
ai_file_analysis.contract_value = "$50,000"
ai_file_analysis.renewal_date = "2025-01-15"

​
AI Agents in Automations

Deploy AI agents within your automation workflows to handle complex interactions and decision-making processes.

​
Giving an Agent a Task

AI agents can be assigned specific tasks within automation workflows, allowing them to handle complex interactions that require understanding, reasoning, and multi-step processing.

​
Task Assignment Process

  1. Context Preparation: Gather relevant data and context for the agent
  2. Task Definition: Clearly define what the agent should accomplish
  3. Agent Execution: The agent processes the task using its capabilities
  4. Result Integration: Use the agent’s output in subsequent automation actions

​
Example: Customer Support Agent Task

Support Ticket Created → Gather Customer Context → Assign Agent Task → Process Agent Response → Update Ticket
Agent Task Configuration:
  • Input Data: Customer history, ticket details, product information
  • Task Instructions: “Analyze the customer issue and provide a recommended solution with steps”
  • Expected Output: Structured response with solution steps and escalation recommendations

​
Agent Task Variables

When an agent completes a task, it provides structured output: 
agent_task.solution_steps = ["Step 1: Verify account", "Step 2: Reset password", "Step 3: Test access"]
agent_task.escalation_needed = false
agent_task.estimated_resolution_time = "15 minutes"
agent_task.confidence_score = 0.89

​
Agent Conversation Ended Trigger

The Agent Conversation Ended trigger captures the completion of agent interactions and makes the conversation context available for further automation processing.

​
Trigger Activation

This trigger fires when:
  • An AI agent completes a conversation with a user
  • The conversation reaches a natural conclusion
  • The agent determines the interaction is complete

​
Available Context

When the trigger fires, it provides access to:
  • Complete conversation history
  • Agent decisions and recommendations
  • User satisfaction indicators
  • Conversation metadata (duration, topics, resolution status)

​
Example Usage

Agent Conversation Ended → AI Summarization → Create Follow-up Tasks → Update Customer Record → Send Satisfaction Survey
Use Cases:
  • Generate conversation summaries for record keeping
  • Create follow-up tasks based on agent recommendations
  • Update customer profiles with interaction insights
  • Trigger post-conversation workflows

​
Conversation Context Variables

conversation.summary = "Customer password reset completed successfully"
conversation.resolution_status = "Resolved"
conversation.satisfaction_score = 4.5
conversation.duration_minutes = 8
conversation.topics = ["password reset", "account security"]
conversation.next_actions = ["Follow up in 24 hours", "Send security tips"]

​
AI Action Patterns

​
Sequential AI Processing

Chain multiple AI actions for comprehensive data processing: 
Document Uploaded → AI File Analysis → AI Classification → AI Summarization → Transform Data with AI → Create Records

​
Conditional AI Logic

Use AI results to make automation decisions: 
Customer Email → AI Classification → 
  IF category = "Complaint" → AI Summarization → Route to Manager
  OTHERWISE → Standard Response → Route to Support Team

​
Agent-Assisted Workflows

Combine AI agents with automation actions: 
Complex Issue Identified → Assign Agent Task → Agent Conversation Ended → AI Summarization → Create Action Plan → Notify Team

​
Best Practices

​
Data Quality for AI Actions

Provide Context: Include relevant background information in AI action inputs Use Clear Variables: Name variables descriptively for better AI understanding Validate Outputs: Check AI action results before using them in subsequent actions

​
Performance Optimization

Batch Processing: Group similar AI operations when possible Selective Analysis: Use conditions to apply AI actions only when necessary Result Caching: Store AI results in variables for reuse within the same automation

​
Error Handling

Confidence Thresholds: Use AI confidence scores to validate results Fallback Logic: Provide alternative actions when AI analysis fails Human Review: Route low-confidence results to human reviewers

​
Integration Examples

​
Customer Support Automation

Complete customer support workflow with AI integration: 
Email Received → AI Classification → 
  IF category = "Technical Issue" → 
    AI File Analysis (attachments) → 
    Assign Agent Task → 
    Agent Conversation Ended → 
    AI Summarization → 
    Create Knowledge Base Entry

​
Document Processing Pipeline

Automated document processing with AI analysis: 
Document Uploaded → AI File Analysis → Transform Data with AI → AI Classification → 
  IF type = "Contract" → Create Contract Record → Set Renewal Reminders
  IF type = "Invoice" → Create Invoice Record → Route for Approval

​
Lead Processing Workflow

Intelligent lead qualification and routing: 
Lead Form Submitted → Transform Data with AI → AI Classification → 
  IF quality = "Hot" → Assign Agent Task → Immediate Follow-up
  IF quality = "Warm" → Add to Nurture Campaign
  IF quality = "Cold" → Add to Long-term Follow-up

​
Monitoring AI Actions

​
Performance Metrics

Track AI action effectiveness:
  • Accuracy: How often AI classifications match expected results
  • Confidence Scores: Average confidence levels across AI actions
  • Processing Time: Time taken for AI analysis
  • Success Rate: Percentage of successful AI action executions

​
Quality Assurance

Regular Review: Periodically review AI action outputs for accuracy Feedback Loop: Use human corrections to improve AI performance Version Tracking: Monitor changes in AI model performance over time

​
Next Steps

Automation System

Learn how to build complete automation workflows with AI actions

Actions Reference

Explore all available automation actions and their configurations

Agent Architecture

Understand how AI agents work and integrate with automations

Best Practices

Follow proven patterns for implementing AI-powered workflows
⌘I