AI is becoming part of many web applications, but integrating AI into a real Drupal application involves more than simply calling an API.
In this session, I will share a practical approach to adding AI capabilities to Drupal applications. We will first look at how services such as OpenAI can be integrated with Drupal for common use cases such as content analysis, classification and automation.
I will then explain when an external AI API may not be enough and how we can train our own model using application-specific data. Using a real-world example, I will walk through preparing and labelling training data, training a text-classification model, testing its accuracy and connecting the trained model with Drupal through APIs.
The session will also discuss some of the challenges we faced in a production application, including incorrect classifications, spelling variations, improving training datasets and deciding when to use an existing AI service versus a custom-trained model.
The aim of this session is not to go deep into AI mathematics. Instead, attendees will get a practical understanding of how AI can be introduced into an existing Drupal application step by step.