Recently, I started exploring the Generative AI capabilities available in Oracle APEX. As part of my learning, I wanted to see how easily I could connect an external AI model to APEX and test it in a real environment.
For this experiment, I chose Google Gemini and configured it through the Generative AI Services available in Oracle APEX.
The setup was quite straightforward, and I wanted to document the steps I followed and a few things I learned along the way.
Why I Tried This
I've been working with Oracle APEX and wanted to understand how AI can be brought into low-code application development.
Instead of building a separate application or writing a custom integration from scratch, APEX provides configuration options for connecting to generative AI services.
That made it a good opportunity for me to try the integration myself and understand what happens behind the scenes.
What I Used
For my setup, I used:
Oracle APEX
Google Gemini
Gemini API key
APEX Generative AI Services
APEX Workspace Administrator access
I also used my own APEX environment to test the configuration.
Configuring Gemini in Oracle APEX
I started by navigating to:
Workspace Utilities → Generative AI Services → Create/Edit
From there, I created a new Generative AI Service and selected Google Gemini as the AI provider.
I gave the configuration a meaningful name so that it would be easy to identify later.
Configuring the API Details
The next step was configuring the Gemini API details.
For the API endpoint, I used:
https://generativelanguage.googleapis.com/v1beta
I then provided the required model information and configured the credentials.
One thing I found useful here was that APEX provides the configuration directly from the UI, so I didn't have to write a separate REST integration just to establish the connection.
Adding the Gemini API Key
The next part was adding the Gemini API key.
I enabled the option to create credentials and entered the API key generated for my Gemini environment.
I also made sure that the API key had the required access for the Generative Language API.
For security reasons, I would recommend never publishing your actual API key in screenshots, blog posts, GitHub repositories, or other public places.
Enabling Gemini for APEX
After configuring the credentials, I looked at the APEX settings.
There are options such as:
Used by App Builder
Default for New Apps
For my testing, I enabled the appropriate option so that the configured AI service could be used by APEX App Builder.
This was one of the interesting parts of the setup because the integration is handled through the APEX configuration rather than requiring me to build the entire connection manually.
Testing the Connection
Before considering the configuration complete, I used the Test Connection option.
I think this is an important step because it immediately tells you whether the basic configuration is working.
If something goes wrong, I would first check:
API key
Model ID
Base URL
Endpoint URL
Credentials configuration
Network/connectivity
Whether the selected model is supported by the configured APEX integration
My Configuration Screen
This is the configuration screen from my APEX environment:
The screenshot shows the Gemini provider, API configuration, model information, credentials, and the App Builder settings.
What I Learned
The biggest takeaway for me was that connecting an AI service to APEX doesn't necessarily require a large amount of custom development.
The configuration is largely handled through the APEX interface, which makes it easier to experiment with generative AI capabilities while continuing to work within a low-code environment.
I also learned that small configuration details matter. The API key, model ID, endpoint, and credential settings all need to match the AI service being used.
Where This Could Be Useful
There are several areas where I think this type of integration could be useful in APEX applications:
Generating application content
Assisting developers during application development
Natural-language interactions
Generating SQL or application-related suggestions
AI-assisted application development
Building smarter business applications
The actual use case will depend on the application and the AI capabilities being enabled.
A Small Lesson From My Experiment
One thing I would recommend to anyone trying this for the first time is don't immediately start troubleshooting the entire application when something doesn't work.
Start with the connection itself.
Check the API key, model, endpoint, credentials, and then use Test Connection.
Once the basic connection works, move on to testing the AI features inside APEX.
That approach makes troubleshooting much easier.
Final Thoughts
This was a useful hands-on experiment for me because it helped me understand how Oracle APEX and Generative AI can work together.
I'm still exploring what can be built with this integration, but I think the combination of low-code development and AI has a lot of potential.
My next step is to move beyond the configuration and build an actual business use case with APEX and Gemini rather than just testing the connection.
Have you tried integrating Gemini, OpenAI, or another LLM with Oracle APEX? I'd be interested to hear about your experience and what you built.
Top comments (0)