The Problem

Without automation:

• You update a product price → AI still tells customers the old price
• You publish a new FAQ → Need to manually add it to knowledge base
• You archive a document → AI might still reference outdated info
• Team wastes time manually syncing data

The Solution

With automation enabled:

• Create a record → Automatically added to knowledge base
• Update a record → Knowledge base updates in real-time
• Delete a record → Automatically removed from knowledge base
• Zero manual work required

Automated Lifecycle Management

• Auto-Create: New records matching domain filters are automatically added to collections
• Auto-Update: Record changes sync to knowledge base in real-time
• Auto-Remove: Deleted records or records no longer matching filters are removed
• Auto-Process: RAG pipeline (parse → chunk → embed) runs automatically

Flexible Domain Filters

• Use Odoo's powerful domain syntax
• Support for any Odoo model (products, partners, documents, custom models)
• Multiple filters per collection
• Easy enable/disable per filter

Smart Automation Rules

• Automatically creates 3 automation rules per domain filter:
  • On creation: Adds new matching records
  • On update: Syncs changes and removes non-matching records
  • On deletion: Cleans up knowledge base
• Orphan cleanup (removes documents not in any collection)

Step 1: Add Domain Filters

1. Go to LLM → Knowledge → Collections
2. Open your knowledge collection
3. Navigate to the Domain Filters tab
4. Click "Add a line"
5. Configure:
   • Model: Select the Odoo model (e.g., Journal Entry, Product, Partner)
   • Domain: Define filter criteria (e.g., [('state', '=', 'published')])
   • Active: Toggle ON

Step 2: Enable Automated Sync

1. Navigate to the Automation tab
2. Enable toggles:
   • Automated Sync: ON (enables automatic synchronization)
   • Auto-Process Resources: ON (automatically runs RAG pipeline)
3. The system automatically creates automation rules

What happens behind the scenes:

When you enable automation, the system creates 3 automated actions for each domain filter:

• On creation: New record created → RAG resource created and added to collection
• On update: Record updated → RAG resource synced; if no longer matches domain, removed from collection
• On deletion: Record deleted → RAG resource removed from collection and cleaned up

Step 3: Done!

Your knowledge base now updates automatically. Create, update, or delete records - the AI knows immediately.

E-Commerce Product Catalog

Goal: Keep product catalog in sync with knowledge base

Setup:

• Model: product.product
• Domain: [('is_published', '=', True)]
• Automated Sync: ON

Result: When you publish a product, AI knows about it instantly. Update price? AI quotes new price immediately.

Company Policies

Goal: Keep HR policies current in knowledge base

Setup:

• Model: document.page
• Domain: [('category', '=', 'policy'), ('status', '=', 'active')]
• Automated Sync: ON

Result: Update remote work policy → AI cites new version to employees immediately.

Support Ticket Solutions

Goal: Build knowledge base from resolved tickets

Setup:

• Model: helpdesk.ticket
• Domain: [('state', '=', 'solved'), ('solution_documented', '=', True)]
• Automated Sync: ON

Result: Mark ticket as solved → Solution automatically added to knowledge base for future reference.

Behind the Scenes

When you enable automation, the system:

1. Creates automation rules: 3 rules per domain filter (creation, update, deletion)
2. Monitors data changes: Uses Odoo's base_automation system
3. Processes matches: When a record matches the domain, creates/updates RAG resource
4. Handles non-matches: When a record no longer matches, removes from collection
5. Runs RAG pipeline: Automatically parses, chunks, and embeds new content

Example Flow

User creates product "Model X" with state=published
↓
Automation rule: "On creation" triggers
↓
System checks domain filter: [('state', '=', 'published')] → MATCH
↓
Creates llm.resource for "Model X"
↓
Adds to knowledge collection
↓
Auto-processes (parse → chunk → embed)
↓
AI can now answer questions about "Model X"

Automation Rule Format

For each domain filter, the system creates:

# On Creation
{
    'name': 'RAG: Collection Name - Model Name (On creation)',
    'model_id': domain_filter.model_id,
    'trigger': 'on_create',
    'state': 'llm_update',
    'llm_collection_id': collection.id,
    'filter_domain': '[]',  # No filter on create
    'llm_auto_process': True,
}

# On Update
{
    'name': 'RAG: Collection Name - Model Name (On update)',
    'model_id': domain_filter.model_id,
    'trigger': 'on_write',
    'state': 'llm_update',
    'filter_domain': domain_filter.domain,
    'llm_auto_process': True,
}

# On Deletion
{
    'name': 'RAG: Collection Name - Model Name (On deletion)',
    'model_id': domain_filter.model_id,
    'trigger': 'on_unlink',
    'state': 'llm_update',
    'filter_domain': domain_filter.domain,
}

API

Sync automation actions programmatically:

collection = env['llm.knowledge.collection'].browse(collection_id)
collection.sync_automated_actions()

Enable automation via code:

collection.write({
    'automated_sync': True,
    'auto_process_resources': True,
})

Automation not working

1. Check domain filter is Active
2. Verify Automated Sync is enabled
3. Check automation rules were created (Automation tab)
4. Test domain filter matches expected records

Resources not processing

1. Verify Auto-Process Resources is enabled
2. Check RAG pipeline configuration (parser, chunker, embeddings)
3. Review system logs for errors

Duplicate resources

1. Check if multiple collections have overlapping domains
2. Review automation rules for conflicts
3. Use "Sync Automated Actions" to rebuild rules