Your Company Knowledge Is Trapped

Scattered across PDFs, Word docs, spreadsheets, images, text files, and folders...

Endless Searching

"Where was that policy document? Was it in SharePoint? Email? The old server?"

Format Chaos

PDFs, .docx, .txt, .xlsx, images - each needs different tools to search

Wasted Hours

Your team spends hours each week hunting for information they KNOW exists

Information Silos

Knowledge locked in different systems - no single place to find answers

Result: Slower decisions, frustrated teams, and knowledge that might as well not exist

There's a Better Way

Transform ALL your documents into unified, AI-searchable knowledge

AI-Powered Knowledge Management

Cut 'Where Is That Document?'
Time to Zero

Your employees spend hours each week hunting through files, emails, and SharePoint. Let AI search your entire knowledge base for them—just ask a question in plain English.

Powered by Retrieval-Augmented Generation (RAG) - the technology that makes AI truly understand your business

Instant Answers
100% Private
Source Citations

What is RAG and Why Should You Care?

Transform your AI from a know-it-all to a know-YOUR-business expert

Think of RAG as giving your AI assistant a perfect memory of YOUR company's knowledge. Instead of getting generic AI answers, your assistant reads your actual documents, policies, and procedures before responding—like having an expert employee who never forgets anything and can instantly access every document you've ever created.

This means accurate, company-specific answers based on YOUR data, not internet generalizations

Transform How Your Team Works

Instant Expert Answers

Get accurate responses based on YOUR company's documentation, not generic internet knowledge.

Save Hours Daily

Stop digging through folders. Ask questions naturally and get the exact information with sources.

Always Up-to-Date

Upload new docs anytime. Your AI instantly knows about changes. No retraining needed.

100% Data Privacy

Your documents stay in YOUR Odoo. Complete data sovereignty, no external sharing.

See the Difference in Action

Real scenarios where RAG transforms daily operations

Employee Onboarding

Without RAG

New sales rep spends days reading manuals, repeatedly asks colleagues about product specs and policies

With RAG

Asks AI: "What are our Pro subscription features?" Gets instant, accurate answers from training materials

"Based on Product Guide v2.3: Pro includes unlimited users, API access, priority support..."

Customer Support

Without RAG

Agent puts customer on hold, searches multiple systems, might give incomplete answer

With RAG

Types question, instantly gets comprehensive answer from manuals and past tickets

"Per Troubleshooting Guide section 4.2 and ticket #1823: Reset procedure is..."

Policy Questions

Without RAG

Manager emails HR, waits for response, might get outdated information

With RAG

Asks: "Remote work policy for international employees?" Gets immediate, accurate answer

"HR Handbook 2024, page 47: International remote work requires VP approval..."

Simple as 1-2-3-4-5

Get intelligent answers from your documents in minutes

1

Upload Documents

Add PDFs, manuals, policies to organized collections

2

Auto Processing

System chunks and indexes content intelligently

3

Ask Questions

Type questions in plain language naturally

4

Get Answers

AI provides comprehensive answers with sources

5

Keep Learning

Add new docs anytime to expand knowledge

Complete Setup Guide

From zero to production RAG in 6 steps with real examples

1

Setup Vector Store

Install and configure Qdrant for vector storage

Install Vector Store Module

Choose a vector store to enable semantic search capabilities. This example uses llm_qdrant, but you can also use:

  • llm_qdrant - High-performance vector database (shown in this guide)
  • llm_pgvector - PostgreSQL extension (if you prefer SQL)
  • llm_chroma - Lightweight embedded database
Vector Store List

Navigate to LLM → Stores to view configured vector stores

Qdrant Configuration

Configure Qdrant connection: host, port, API key, and collection name

Tip: You can run Qdrant locally with Docker: docker run -p 6333:6333 qdrant/qdrant
2

Configure Embedding Provider

Setup OpenAI for text embeddings

Install Module: llm_openai

Configure OpenAI provider and select the text-embedding-3-small model for efficient, cost-effective embeddings.

OpenAI Model Configuration

Select text-embedding-3-small model for document embeddings

Alternative: You can also use llm_mistral for Mistral embeddings or llm_ollama for local embeddings

BONUS: Parse Images with AI Optional

Extract text from images, receipts, screenshots with Mistral OCR

Install Module: llm_knowledge_mistral

Want to extract text from images, receipts, screenshots, or handwritten notes? Install the Mistral parser extension to unlock OCR capabilities!

Mistral OCR Models

Available Mistral models with OCR/vision capabilities

Sample Grocery List Image

Sample image: Handwritten grocery list

Mistral Parser Results

Perfect parsing! Mistral extracted all text from the handwritten image

What Can You Parse?
  • Receipts & Invoices: Extract line items, prices, totals
  • Handwritten Notes: Convert handwriting to searchable text
  • Screenshots: Pull text from app screenshots, charts
  • Scanned Documents: OCR any scanned paper document
  • Product Labels: Read ingredients, specifications, warnings
How It Works:
  1. Upload image files (.jpg, .png, .webp, .gif) to your collection
  2. Select Mistral OCR model as parser
  3. Process resources - Mistral extracts all text automatically
  4. Text is chunked, embedded, and made searchable like any other document
3

Create Knowledge Collection

Organize documents into searchable collections

Create a collection that links your vector store (Qdrant) with your embedding model (OpenAI). Collections organize related documents together.

Collection Creation

Create collection: Select vector store and embedding model

Key Settings:
  • Vector Store: Qdrant (from Step 1)
  • Embedding Model: text-embedding-3-small (from Step 2)
  • Chunk Size: 512-1024 tokens (adjustable)
4

Upload & Process Documents

The complete RAG pipeline in action

Step 4a: Upload Documents

Add documents to your collection in multiple ways:

  • Upload Files: PDFs, text files, markdown, images with text
  • External URLs: Provide links to download files from external sources
  • Website Links: Fetch and parse content directly from web pages
Upload Wizard

Upload documents via file picker or URL

Step 4b: View Uploaded Resources

See all documents in your collection and trigger processing manually or automatically.

Resources List

Resources list showing uploaded documents ready for processing

Step 4c: Processing Pipeline

Watch the magic happen: Parse → Chunk → Embed

Processing Pipeline

Real-time processing status: parsing text, chunking content, generating embeddings

Step 4d: View Chunks

Inspect individual chunks with their embeddings, ready for semantic search.

Chunks View

Chunk list showing embedded text pieces with metadata

Success! Your documents are now searchable with AI-powered semantic search
5

Use with LLM Assistants

Query your knowledge directly in Odoo

Install Module: llm_tool_knowledge

Enable your LLM assistants to search your knowledge collections. Create an assistant with knowledge retrieval tools.

LLM Assistant Chat

Chat with your LLM assistant that can search and cite your documents

Example Queries:
  • "What are our company's remote work policies?"
  • "Show me the product specifications for Model X"
  • "How do I reset a customer's password?"
6

Connect External Apps

Use with Claude Desktop, Cursor, Ollama, and more

Install Module: llm_mcp_server

The Model Context Protocol (MCP) server exposes your knowledge collections to any MCP-compatible application.

Claude Desktop Usage

Claude Desktop app querying your Odoo knowledge base with source citations

Compatible Apps: Cursor, Claude Code, Claude Desktop, Codex CLI, and many more MCP-compatible applications

Complete! Your Odoo knowledge is now accessible from any AI application that supports MCP

What's Next?

Auto-Sync

Install llm_knowledge_automation to automatically keep collections synced with your data

Alternative Stores

Try llm_pgvector (PostgreSQL) or llm_chroma for different vector storage options

Advanced Parsing

Use llm_mistral for image/OCR parsing or llm_knowledge_llama for LlamaIndex chunking

Enterprise-Ready Knowledge Management

Powerful features designed for real business needs

Organized Collections

Create separate knowledge bases for different departments or topics

  • HR Policies
  • Product Documentation
  • Technical Manuals
  • Sales Playbooks

Intelligent Chunking

Documents automatically split into optimal pieces for accurate retrieval

Preserves context while maximizing search accuracy

Semantic Search

Finds relevant content even when exact words don't match

Ask "vacation policy" → finds "PTO guidelines"

Multiple Formats

Process various document types seamlessly

  • PDF documents
  • Text files
  • Markdown files
  • Web pages (HTTP)

Flexible Storage

Choose your preferred vector database

  • PostgreSQL pgvector
  • ChromaDB
  • Qdrant
  • More coming soon

Source Citations

Every answer includes references to source documents

Build trust with transparent, verifiable responses

Seamless Integration

Works perfectly with the entire Odoo LLM ecosystem

AI Assistant Integration

Enhance your LLM assistants with your company's knowledge base

Works with:
  • LLM Assistant - Create knowledge-aware AI assistants
  • LLM Thread - Chat with document context
  • LLM Tool Knowledge - RAG tools for any model

Provider Flexibility

Use any LLM provider for embeddings and generation

Supported providers:
  • OpenAI - GPT models and embeddings
  • Mistral - European AI models
  • Ollama - Local, private models
  • More providers available

Transform Your AI Into a Company Expert

Stop settling for generic AI answers. Give your team instant access to YOUR knowledge.

Install in Minutes
100% Private
Instant ROI

Documentation & Support

Everything you need to succeed

Complete Documentation

Step-by-step guides and best practices included

GitHub Repository:

github.com/apexive/odoo-llm

Full documentation, examples, and tutorials

Open Source

100% open source with active community support

License: LGPL-3

Customize, extend, or contribute back to the community