AI Services

Hi

PART 1: The DREAM

We have an application called DREAM “Database Risk Evaluation & Analysis of Maturity” (you can watch the Funnel page of DREAM here: https://it-crowd.funnels.mastertools.com) - that evaluates the database security posture from eight different areas by asking the participant five questions for each area. This APP is developed using APEX 24.1 on and Oracle 23ai ATP Database in the OCI.

• you can change the language to English in the browser; just watch the video how the app works.
• if you want to give a try you are most welcome to give an attempt and your comments: https://b1r1mryygmx24ne-encuestas.adb.us-ashburn-1.oraclecloudapps.com/ords/r/dreamer/encuesta/validar-e-mail

At the end the application we give a preliminary analysis of the current database security posture indicating the risk-score as well as the maturity-score in a tabular form as well as a Spider-Chart.

Part 2: The AI

PARALLELLY - We made a collection of PDF documents (product datasheets, whitepapers, tech briefs, reports etc.) related to the necessary hardware and software solutions that can improve the security posture for the databases. And using the “_Load + Split + Embed + Vectorize_” cycle of RAG, we have created a ChromaDB with the Vectorized embeddings of these documents. (first part of the Demo by Hemant Gahankari).

• This gives us the flexibility to load and maintain the PDFs, select a choice of LLM Models for generating the vectors etc.
• However to be able to use this RAGged info along with the “context” of a client we need to do some hard coding using python and all the AI dependencies and to host a StreamLit app

And here comes the need to converge DREAM with AI…

Part 3: The DREAM-AI

As you can see, so as to save the valuable CPU cycles of the ATP database we offloaded the “_document crunching and vector creation_” to a separate VM. However we would like to insert these Vectors into the VECTOR Column of a table in the Oracle 23ai ATP Database where our APEX App is residing.

In this way, instead of coding a separate AI Chatbot with the VectorDB outside of our APEX based app, we can use the APEX + AI Integration and the Clients can readily consult the Chatbot regarding their current security posture, consequences, recommendations etc. based on the PROMPT that is based on the Answers the client has answered as well as the Vectorized RAG Info inserted into a table in our APEX App.

Question:

• How to Load the RAG “embeddings + vectors” from non-Oracle databases (FAISS, ChromaDB, MySQL HeatWaver etc.) into a table with VECTOR column in Oracle 23ai Database?

Regards