Chains — Workflow Automation

Chains is a fully integrated workflow automation engine for ELONIQ. With it, you can:

• Model complex multi-step processes visually
• Automate document routing and processing between systems
• Integrate AI-based classification and extraction
• Configure error handling and retries
• Trigger workflows via events (file upload, webhook, schedule)
• Orchestrate multiple DMS/ERP systems (ELO, SharePoint, databases, REST APIs)

Core Features

• Visual Editor — Drag-and-drop node editor with live preview
• 10+ Node Types — Action, Decision, Loop, AI, Parallel, Try/Catch, Delay, User-Task, etc.
• Module Integrations — Direct actions for elobase, eloimporter, documentclassifier, database, and more
• AI Tools — LLM-powered text classification, extraction, summarization
• Variables & Templating — Global variables, {{dot-notation}}, data flow tracking
• Error Handling — Retry logic, Try/Catch blocks, conditional branches
• Monitoring — Complete run history, logs, performance metrics

Node Types

• Trigger/Start — Initiates a chain run
• Action — Executes module-specific operations
• Decision — Conditional branching (if/else)
• Loop — Iteration over arrays/results
• Parallel — Concurrent execution of multiple branches
• AI — Claude LLM with tools and prompts
• User Task — Manual approval/input
• Delay — Waits for a specified time
• Try/Catch — Error handling with fallback
• End — Terminal node

Available Module Actions

Each module registers specific actions. See Actions Reference below.

Step 1: Create Chain

1. Navigate to Automation > Chains
2. Click New Chain or From Template
3. Enter name and description
4. Click Create

Step 2: Build Chain Logic

1. Click Edit to open visual editor
2. Add a Start Node if not present
3. Click Add Node and select type + module
4. Configure parameters (template variables supported: {{variable}})
5. Connect nodes with edges — drag from output to input

Step 3: Configure Triggers

1. Open Triggers tab in editor
2. Click Add Trigger
3. Select type (File Watcher, Webhook, Schedule, Manual)
4. Configure conditions (e.g. file patterns, cron)

Step 4: Test

1. Click Test to run a test execution
2. Review logs and variables in run history
3. Iterate until behavior matches expectations

Step 5: Activate & Monitor

1. Mark chain as Active
2. Click Save
3. Monitor executions in Run History

• Small, focused chains — Break large workflows into smaller chains; use invoke_chain AI tool to orchestrate
• Error handling — Use Try/Catch or Decision nodes to handle errors. Never silently ignore failures.
• Logging — Use log actions to store debug information. Helps troubleshoot later.
• Variable naming — Use descriptive names: source_file, elo_obj_id, dc_fields. Follow conventions.
• Timeouts — Set timeouts on long operations (classification, upload): timeout: "5m"
• Retry strategies — Use retry for transient errors (network, timeouts), not for logic errors
• Variable expansion in paths — Avoid hardcoded paths; use {{variable}} substitution for flexibility
• Test before production — Test with real (but test) documents before activation

Example 1: File Import + Classification + Archive

Start (File Watcher: *.pdf in /inbox)
  ↓
classify_manual_fallback Chain
  ├→ documentclassifier.classify
  └→ Decision: dc_confidence >= 0.9?
       ├→ [Yes] elobase.upload → elobase.update_metadata
       └→ [No] send_mail ("Manual review required")

Example 2: ERP Data Enrichment

Start (Webhook: /chain/enrich)
  ↓
database.query (SELECT * FROM customers WHERE id = {{customer_id}})
  ↓
elobase.update_metadata (with db_result_first fields)
  ↓
send_discord (Status: "Successfully enriched")

Example 3: Parallel Processing

Start (Webhook: /export)
  ↓
Parallel Node:
  ├→ Branch 1: eloexporter.export_document → store PDF
  ├→ Branch 2: elometadata.update_from_source → sync ERP
  └→ Branch 3: database.query → update log table
  ↓
End ("All branches completed")