| Availability |
Odoo Online
Odoo.sh
On Premise
|
| Odoo Apps Dependencies |
Discuss (mail)
|
| Lines of code | 7391 |
| Technical Name |
support_genie_ai |
| License | LGPL-3 |
SupportGenie AI
Cognitive AI · Multi-LLM · RAG · Omnichannel · Copilot
SupportGenie AI transforms your customer support with a full cognitive AI platform built natively for Odoo 19. Connect 6 AI providers, power a semantic RAG knowledge base, automate triage across 9 channels, and give your agents an intelligent copilot â all in one module.
ð¥ What's New in v19.0.3
- RAG Retrieval Fix: Semantic query normalization expanded â "vacation days" â "paid leave days", lower threshold (0.45) for better HR policy matching, per-result diagnostics logging, and a decision tree that never falls back to General Knowledge when KB documents exist.
- Demo Knowledge Bases: Pre-installed Employee Leave Policy, Password Reset Guide, and Subscription & Refund Policy â RAG works out of the box with zero configuration.
- Portal Redesign: Bootstrap-first ticket detail view with responsive grid, status badges, AI response cards, and Google Translate compatibility fix.
- Improved Fallback UX: When the KB is still indexing or retrieval confidence is low, the LLM now says so transparently instead of pretending no company policy exists.
SupportGenie AI transforms your customer support with a full cognitive AI platform built natively for Odoo 19. Connect 6 AI providers, power a semantic RAG knowledge base, automate triage across 9 channels, and give your agents an intelligent copilot â all in one module.
ð¸ Screenshots
AI Portal Dashboard
view and manage AI tickets
Ticket Detail
AI-powered response with knowledge sources
Knowledge Base
RAG-powered semantic search with pgvector
AI Configuration
multi-provider LLM settings
⨠Key Features
ð¤ AI & LLM
- Multi-provider support: OpenAI, Anthropic, Gemini, DeepSeek, Mistral, OpenRouter
- Automatic provider fallback on failure
- Configurable temperature, token limits, and per-provider cost tracking
- Multilingual AI responses (EN, ES, FR, DE, AR, HI)
ð RAG Knowledge Base
- Semantic search powered by pgvector embeddings
- Import wizard for bulk knowledge loading
- Configurable chunk size, overlap and similarity threshold
- Automatic nightly re-indexing via cron
- Demo knowledge bases pre-installed â RAG works out of the box
ð¯ Intent & Categorisation
- Automatic intent detection (Refund, Technical, Sales, Complaint, â¦)
- Smart category learning from agent corrections
- Confidence-based routing: auto-respond, manager review, escalate
ð Omnichannel
- Email, Website Chat, WhatsApp, Telegram
- Facebook Messenger, Instagram, Manual & API
- Unified ticket view regardless of source channel
ð¡ï¸ Enterprise Security
- Prompt injection detection with configurable regex rules
- PII detection and masking before LLM calls
- Per-user rate limiting with audit trail
- Full audit log: LLM calls, token usage, security events
ð Reporting & SLA
- AI-predicted SLA resolution time and breach alerts
- Enterprise dashboard with OWL + Chart.js
- Feedback learning loop with accuracy tracking
- Customer memory engine for context-aware responses
âï¸ Technical Requirements
- Odoo version: 19.0
- Python packages:
openai,langchain,langchain_openai,langgraph - PostgreSQL extensions:
pgvector(auto-created on install),pg_trgm - Optional:
numpyfor local fallback embeddings when no provider is configured
ð Getting Started
- Install the module â
pgvectorandpg_trgmare created automatically. - Go to SupportGenie AI â Configuration â AI Providers and add your API key for at least one provider.
- In Configuration â AI Settings select your primary (and optional fallback) provider.
- Upload knowledge articles via Knowledge Base â Import Wizard, or use the pre-installed demo knowledge bases.
- Start creating tickets â the AI will analyse, categorise, and suggest responses automatically.
- Enable the Portal to let your customers view AI tickets, ask questions, and get instant answers.
ð¬ Support
For questions, bug reports, or feature requests please contact prashantyadav195@gmail.com.
Please log in to comment on this module