Hi everyone. I’d like to learn how people are currently implementing Artificial Intelligence in NetSuite, particularly integrations with OpenAI or other AI models. For use cases such as an AI assistant that can query customer information, products, sales, inventory, etc., what architecture are you using? I have a few questions: • Is the integration typically done directly within NetSuite using SuiteScript/REST, or do you recommend having an intermediate backend? • If you want the AI to query information from NetSuite, is it necessary to use embeddings + a vector database (RAG), or are there cases where you simply query NetSuite using SuiteQL/REST and provide the relevant information to the AI model? • How are you handling security and permissions to ensure the AI cannot access or modify information it shouldn't? • What architecture or approach has worked best for you in production? I’m particularly interested in learning from real-world experiences and best practices for integrating OpenAI with NetSuite.

We initially did a POC where the incoming Emails which create Cases are passed to OpenAI via a netsuite UE script to our Items model OpenAIs API. We have a structured prompt asking if it's a quote / sales order request and if so pull out the items and required quantities, emails, company name, PO Number. In return we get a JSON object which we build into an unapproved quote or sales order for sales rep review. Instead of sales reps reading emails and manually creating the transaction, they are reviewing what has been created. This is now in Production for the past two months.