Integrating Jiwa ERP with SQL and AWS bedrock

[amyacker]

Hi,

I'm working on a project involving Jiwa ERP, Microsoft SQL Server, and AWS Bedrock, and I'd appreciate some guidance from anyone who has implemented something similar.

Jiwa ERP stores product, inventory, pricing, and customer related information in SQL Server. We'd like to use AWS Bedrock to build an AI assistant that can answer natural language questions such as: "Which products are currently in stock?" "Show me construction equipment under SAR 5,000." "Which suppliers have delivered this item in the past six months?" "Recommend alternative products if an item is out of stock."

My questions are:

What is the recommended architecture for securely exposing Jiwa SQL data to AWS Bedrock without allowing the model direct database access?
Are you using a Retrieval-Augmented Generation (RAG) approach, or are you generating SQL queries from user prompts?
How do you keep AI responses synchronized with frequently changing inventory and pricing data?
Have you encountered performance or security challenges when connecting SQL Server data with Bedrock?
What techniques have worked well for handling permissions so users only see data they are authorized to access?

If anyone has experience integrating Jiwa ERP using AWS Bedrock with SQL Server in an ERP environment, I would really appreciate hearing about your architecture, lessons learned, or best practices.

Thanks in advance.