Skip to Main Content

Build Context-Aware Chatbots with Oracle APEX and OCI Generative AI Agents

Question And Answer

  • #SELECTION#
    #ICON_HTML#

    "Is API calling enabled only for specific regions or tenancies? I can’t see this feature in all tenancies within the same region (FRA)."

    OCI’s AI Agents are not yet available in every region. As of now, they’re only offered in specific Oracle Cloud regions;Refer to this to learn more: https://docs.oracle.com/en-us/iaas/Content/generative-ai-agents/overview.htm#regions

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Can I create this in free autonomous OCI Console and APEX

    Absolutely, with every OCI Tenancy you can create 2 ADB instances with which you will get APEX pre-configured

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Can the LLM or knowledge base hold memory of the PDF document? like I have some project information, and I would upload a newer version of it , if I ask the bot , about the addendum to the document compared to original, how would it handle it?

    If you are uploading the updated document as well from the APEX application, then it initiates the ingestion job which will have your updated vectors as data to answer you

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Can this entire list of questions and answers be sent along with the recording? I find it very useful to go through it after the zoom, as for now cannot focus on both question and demo.

    The Q&A will be available on https://apex.oracle.com/officehours/ for this session in the upcoming days.

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Can we make this ingestion dynamic? as the docs change, will it be able to vectorize new information?

    Here the file is being stored in OCI Object Storage and Object storage has a way to raise an event upon change, and you can then invoke your ingestion job dynamically. But in this demonstration, you are uploading the doc via APEX, so the job will be triggered automatically as configured by Roopesh

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Can you repost the URL to subscribe to the APEX newsletter?

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Could you please elaborate on Injestion job creating embeddings?

    Yes, in an OCI-based Generative AI setup, the ingestion job is responsible for creating embeddings from your source documents before storing them in the OCI Vector Store.

    It store them in the associated Knowledge Base for vector-based retrieval and analysis.

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Currently we cannot configure GenAI Agents in APEX AI Configuration section. Is this in roadmap of next releases?

    The configuration section in Oracle APEX is for configuring Generative AI service providers and not for Gen AI Agents as of now.

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Do we need to have a 23ai version of the database , can this not be implemented using 19c ?

    For this particular solution being demonstrated, 19c would do as vectors are not stored in the database.

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Does the RAG chatbot memory last only for that particular user session or does it save it as its longterm memory?

    It is persisted per session unless you store the request and response in a table as shown.

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Does this mean the page gets reloaded after each prompt?

    No, It is calling a dynamic action to call OCI Gen AI Agents chat API

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    For this demo.... I need provisiong OCI Generative Agent AI and any LLM?? or only Agent AI?

    This demonstration is to leverage OCI Generative AI Agent. So, you need OCI Gen AI Agent

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Hi APEX Team, In APEX, I have created a chatbot application. I am calling Gen AI by using a Flask URL, storing the response, and displaying it using an interactive grid. But sometimes its response is in a tabular format, and I am not able to beautify that and display it in tabular format.

    You can detect if the response looks like Markdown or table and convert it into array of objects and then parse it to be able to display as columns in grid

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    In shared components > AI configurations we can set RAG sources manually. Is there a supporting APEX package/api to manage RAG sources?

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Is this sending full, binary PDF documents to the LLM that it decides are relevant? That could get large.

    The OCI Gen AI Agents will break your entire doc/pdf to chunks and store them in Object storage, when you ask a question that will also be broken to vector chunks and sends the vectors to LLM rather than the document

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    OCI Gen AI does it support MCP , if so does Oracle APEX support any MCP natively

    OCI Generative AI complements MCP, especially in the context of vector search and large language models (LLMs) over structured and unstructured content. While Oracle APEX does not natively expose MCP as a feature, it leverages many of its capabilities through the underlying Oracle Database.

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Where do we download the Lab for this?

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    Where do you define the source as Object Storage? Can we ask it to take from a Table in our schema? Also taking from Confluence is possible?

    That is done while configuring the gen ai agents in oci and yes you can choose from Object storage or opensearch or Oracle Database 23ai

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    what's SKU require for this ?

    OCI Generative AI Agents

    #MISC#
    #ACTIONS#
  • #SELECTION#
    #ICON_HTML#

    which vector store is being used?

    while configuring the OCI Gen AI Agent you will be choosing the Knowledge base data store type. So your vector store depends on what you choose there

    #MISC#
    #ACTIONS#

Featured Speakers

Workshop Info

Session Has Completed - 24 July 2025
1 Hour
English
Oracle APEX AI Application Generator

Hidden Translations

Join
Share
Register
English
Unregister
Outline
Objectives
Resources
24 July 2025
Session Has Completed - 24 July 2025
Session has been cancelled
You Are A Speaker/Panelist
1 Hour
Pending
Prerequisites
Waitlisted
Rejected
Question And Answer
Register And Join
Add To Calendar
Registration Status
Series Name
Session Time
Log In To Register
You Are Registered. You Will Be Able To Join The Session 15 Minutes Before It Starts.
You Are Signed Up for the Lab
Sign Up for Lab
Remove Lab Sign Up