Overview

Event-driven automation that monitors system events and executes predefined actions when conditions are met.

Elementum’s Automation System is an event-driven framework that responds to activities within your system. Automations monitor system events and execute predefined actions when specific conditions are met.

Automations must be created within an App, Element, or Task. Navigate to Automations under Workflows & Processes in the left navigation menu of the object where you want to build.

How Automations Work

Every automation follows a listen-think-act pattern:

1. Listen (Triggers) - Monitor for specific events happening in your system
2. Think (Conditions & AI) - Evaluate the event and determine what should happen
3. Act (Actions) - Execute the appropriate response automatically

Manual vs. Automated: Customer Support

Manual process: Customer emails support → Email sits in inbox → Agent reads email → Agent searches for customer → Agent creates ticket → Agent assigns ticket → Agent sends confirmation

Automated process: Customer emails support → Email Received trigger fires → AI analyzes email content → Search Records finds customer → Create Record generates ticket → AI Classification determines priority → Make Assignment routes to appropriate agent → Send Email Notification confirms receipt

Triggers: Understanding Events

Events represent activities within your system. Every interaction, data change, or scheduled occurrence can trigger an automation.

Record Events

• Record is Created: New data enters your system
• Record is Updated: Existing data changes
• Approval Process Status Updated: Approval workflows advance

Content Events

• Attachment is Added: Files uploaded to records
• Comment Added: Team members communicate on records

Data Events

• Data Mine: Scheduled, condition-based reactions to CloudLink-backed table data (also the way to schedule an automation)
• Email Received: Incoming emails that need processing

Time Events

• Time-Based: Fire relative to a date field on a record (for example, 7 days before a contract expiration date)
• Survey: Responses submitted through forms

Workflow Events

• On-Demand Trigger: Custom triggers with inputs/outputs for complex workflows
• Agent Conversation Ended: AI agent interactions complete with context

For complete trigger details and configuration options, see the Automation Triggers Reference.

Schedule an automation

To run an automation on a recurring cadence (for example every 15 minutes, hourly, or daily at 8:00 UTC), use a Data Mine. Data Mining is primarily for monitoring CloudLink-backed table data and firing on state transitions, but the same scheduled-evaluation behavior is the only mechanism in Elementum for running an automation on a recurring schedule. See Schedule an automation with a Data Mine for the trigger record pattern.

The Time-Based trigger is not a recurring scheduler. It fires before, on, or after a date field on a specific record (for example, 7 days before a contract’s expiration date). Use Time-Based when timing is tied to record data; use a Data Mine when you need a recurring cadence.

AI Integration

AI actions process information and make decisions within your automation workflows. Each action receives data through variables and produces structured outputs.

• AI Classification - Categorizes unstructured data into predefined groups (e.g., routing support tickets by type)
• AI Summarization - Condenses lengthy content into key insights (e.g., generating executive summaries from customer feedback)
• AI File Analysis - Extracts structured information from documents (e.g., pulling vendor, amount, and due date from invoice PDFs)
• Transform Data with AI - Cleans, normalizes, and enhances data (e.g., standardizing addresses and phone numbers)

Variables control what information AI actions receive and what they return. For example:

Customer Data (variable) → AI Classification → Determines Customer Tier
Contract Document (variable) → AI File Analysis → Key Terms Extracted

Build Your First Automation

1. Choose your trigger

   Select the event that starts the automation. For a customer support workflow, use Email Received to capture incoming requests. See the Triggers Reference for all available triggers and configuration options.

2. Add AI actions

   AI Classification: Analyze the email content and categorize it (Bug, Feature Request, General Support) — output stored as email_category

   AI Summarization: Create a brief summary of the customer issue — output stored as issue_summary

3. Add response actions

   Search Records: Find the existing customer using their email address — output stored as customer_record

   Create Record: Generate a support ticket using the customer record, category, and summary — output stored as support_ticket

   Make Assignment: Route to the appropriate team based on email category — output stored as assigned_agent

   Send Email Notification: Confirm receipt to the customer with the ticket number and assigned agent

   See the Actions Reference for all available actions and configuration options.

4. Review the complete flow

   Email Received → AI Classification → AI Summarization → Search Records → Create Record → Make Assignment → Send Email Notification
   

Automations are saved in Draft status by default (indicated by a gray pill) and will not fire until published. To activate your automation, click Publish — the status pill turns green to confirm it is live. To deactivate a published automation, select the More icon in the top-right corner of the automation configuration page.

Validate Access Before Publishing

Before publishing an automation, run a Validate Only check to confirm that the user (or service account) responsible for running the automation has the roles and data access required for every object the automation touches. This prevents an automation from breaking after publish because the publisher lacks access to a referenced element, table, or related record.

To run a validation check:

1. Open the automation you want to check.
2. In the top-right corner, click the dropdown arrow next to Publish.
3. Select Validate Only.

Elementum scans every action in the automation and lists each object it interacts with, along with the access level detected for the assigned account. Each object is reported as a passed or failed check.

• Passed Checks — The account has sufficient access to the listed object for the automation to run.
• Failed Checks — The account is missing roles or data access for the listed object. Click Edit Access next to the object to update permissions, then re-run the validation.

If you don’t have permission to manage permissions on the target object, Edit Access silently routes you to the home page instead of opening the access settings. Ask an administrator to update the object’s permissions.

Validate Only works for both regular users and service accounts. Running it before each publish — especially when a different user is publishing changes — helps catch access gaps before they cause runtime failures.

Reorder Actions

Reorganize your automation flows using drag-and-drop. Each action has a handle that lets you move it to a new position in the sequence. You can drag individual actions or entire logic blocks (such as If/Else) in one move.

• Real-time validation — The system flags broken variable references or dependencies as you reorder, so you can fix issues before saving.
• Undo — Click Undo to revert a reorder if the new arrangement breaks your flow.

After reordering, review any actions that reference variables from earlier steps to confirm the dependencies are still valid.

Automation Patterns

Decision Tree

Use IF conditions to create branching logic based on record data:

Record is Updated → IF customer tier = "Enterprise"
  → Send Message to Teams (priority channel)
  → Make Assignment (senior agent)
OTHERWISE IF customer tier = "Standard"
  → Send Email Notification (standard response)
  → Make Assignment (general queue)

Data Processing Pipeline

Chain multiple AI actions for complex data transformation:

File Uploaded → AI File Analysis → Transform Data with AI → AI Classification → Create Record → Send Email Notification

Approval Workflow

Combine AI with human decision-making:

Record is Created → AI Classification → IF amount > $10,000
  → Start Approval Process (CFO approval)
  → Send Message to Teams (finance team)
OTHERWISE
  → Update Record Fields (auto-approved)

Variables

Variables carry information between automation actions, connecting outputs from one step to inputs of the next.

Trigger variables come from the initiating event:

• trigger.record_id - ID of the record that triggered the event
• trigger.user_email - Email of the user who caused the event
• trigger.timestamp - When the event occurred

Action variables are outputs from executed actions:

• search_result.customer_name - Customer name from Search Records
• ai_classification.category - Category from AI Classification
• new_record.record_id - ID of newly created record

Use descriptive variable names (customer_priority_level rather than priority), and use IF conditions to validate variable values before passing them to downstream actions.

Actions

Automations support a broad set of built-in actions across several categories. For complete details, examples, and configuration options, see the Automation Actions Reference.

Logic Actions

Use IF conditions to branch behavior and Repeat for Each to loop over collections.

Record Actions

Create, update, search, and relate records. Manage approvals, assignments, and field locking.

Communication Actions

Send email notifications, post Microsoft Teams messages, and add comments to records.

File Actions

Read files, extract structured data with AI analysis, and process ZIP archives.

Data Actions

Run calculations, set variables, classify content, summarize text, and transform data with AI.

External Actions

Call external APIs and execute custom functions for specialized integrations.

Intelligence Actions

Classify, summarize, transform, and search data with AI. Run autonomous agent tasks.

Monitoring and Debugging

Automation Tracing

Automation Tracing gives you a detailed view of each automation run so you can confirm success or diagnose failures.

To access Automation Tracing, click the History icon at the far right of an automation on the Automations page, or from the automation details page.

From the tracing view, you can:

• Review each execution — See start time, duration, automation version, and status for every run.
• Inspect actions within a run — Click an execution to view which actions succeeded or failed.
• Examine action data — Click an individual action to see its returned inputs and outputs.

Use Post Comment actions to log progress at key points in your automation — this makes it easier to trace execution and verify that variables contain expected values at each step. You can also access automation history through App Health.

Version History

Every published change to an automation is tracked as a version. To review past versions:

1. Open the Automations page under Workflows & Processes in the app’s navigation menu.
2. Click the version number in the Version column for the automation you want to inspect.
3. Review the changes made in each version and the user who made them.
4. To revert to an earlier version, click the Restore as Draft icon next to that version.

Restoring a version loads it as a draft. The reverted changes are not live until you publish the automation.

Common Use Cases

Invoice Processing

Scenario: Automatically extract and route invoices uploaded as PDF attachments

Trigger: Attachment is Added (invoice PDF)

Flow:

PDF Uploaded → AI File Analysis → Create Invoice Record → IF amount > $5,000
  → Start Approval Process
OTHERWISE
  → Update Record Fields (auto-approved) → Send Email Notification

Customer Onboarding

Scenario: Welcome new customers and set up their accounts automatically

Trigger: Record is Created (new customer)

Flow:

New Customer Created → AI Classification → Generate Report (welcome packet) → Send Email Notification → Create Record (project) → Make Assignment (account manager)

Sales Lead Qualification

Scenario: Score and route inbound leads based on fit and quality

Trigger: Record is Created (lead form submission)

Flow:

Lead Form Submitted → AI Classification → Update Record Fields (lead score) → IF score > 80
  → Make Assignment (senior sales rep) → Send Email Notification (immediate follow-up)
OTHERWISE
  → Make Assignment (lead nurturing queue)

Project Status Monitoring

Scenario: Alert stakeholders when a project status changes, with escalation for at-risk projects

Trigger: Record is Updated (project status change)

Flow:

Project Status Updated → AI Summarization → Generate Report → Send Message to Teams → IF status = "At Risk"
  → Send Email Notification (escalation)

---

For design principles, performance strategies, and proven patterns, see Automation Best Practices. For complete action details and examples, see the Automation Actions Reference. For trigger configuration and use cases, see the Automation Triggers Reference.