How I Built a RAG-Powered Conversational Assistant for Odoo ERP

Short summary

Full-Stack Developer built a RAG-powered conversational assistant in Odoo ERP using LangChain and pgvector, enabling sales reps to query data naturally instead of navigating multiple screens. Fine-tuned a 7B model with QLoRA on real support tickets, reducing lookup time to seconds. Key insight: domain-specific fine-tuning outperforms larger models at lower inference cost.

• •Built RAG system within Odoo using LangChain + pgvector embeddings and context-aware retrieval
• •Fine-tuned 7B model with QLoRA; outperforms larger models at half the inference cost
• •Reduces enterprise data queries from 4-5 screens to seconds via natural language interface

Generated with AI, which can make mistakes.