Integrating Drupal and AI for Enhanced Customer Service: A Case Study of Alexela.ee
I’m Anuar Alfetahe, the Dev Lead at ADM Interactive. With over five years of experience in software and web development, particularly with Drupal, I’ve had the opportunity to work on some innovative projects. Today, I’m excited to share one such project: the integration of a customer service chatbot using Drupal and OpenAI’s GPT-3.5 Turbo model for Alexela.ee.
Project background
Our journey began with a previous chatbot, entirely handcrafted and configurable within Drupal‘s content management system. However, to enhance user interaction and provide more dynamic responses, we embarked on integrating this existing system with OpenAI’s GPT-3.5 Turbo and GPT 4, the same models powering ChatGPT.
Key Developments and Innovations
Integration with OpenAI’s GPT-3.5 Turbo: We merged our existing chat solution with the advanced capabilities of GPT-3.5 Turbo, allowing for more natural and varied user interactions.
Customizable AI Model: Administrators now have the flexibility to tweak the model’s parameters, enhancing creativity or focusing on more repetitive, informative responses.
Incorporating Alexela-Specific Information: Administrators can input specific information about Alexela into the system, guiding the AI to generate more relevant and accurate responses for users.
Initial Model Tuning and Analysis: We conducted extensive research and experimentation with various AI models, including a deep dive into OpenAI’s documentation. This allowed us to fine-tune the AI to stay aligned with topics relevant to Alexela and avoid veering off into unrelated areas.
Enhanced User Experience: To make the interaction more engaging, we added functionalities allowing users to provide feedback through likes/dislikes and text comments.
Data Analytics: Integration with Google Analytics enables us to track system variables like the source of the model’s information, enriching our understanding of user interactions and preferences.
Front-End and Security Architecture: The chat application, built on the Vue framework, interfaces with Drupal, which then communicates with the OpenAI service. This design ensures that client browsers do not directly interact with OpenAI services, maintaining security and control over the traffic.
Drupal’s Role: Within this architecture, Drupal plays a crucial role in tuning and administering the OpenAI model, collecting user feedback, and controlling and limiting the use of OpenAI services based on IP and session data.
The Chatbot ‘Alex’
Our chatbot, named Alex, begins interactions with a disclaimer about potential inaccuracies due to fluctuating data, setting realistic expectations. Alex is designed to assist with queries about Alexela’s services, products, pricing, customer support, and other related topics, providing detailed information and guidance.
Conclusion
This project exemplifies ADM’s expertise in seamlessly integrating AI with established web technologies like Drupal to enhance customer service experiences. It reflects our commitment to innovation, security, and efficiency, ensuring our solutions are not just technologically advanced but also user-centric and practical.
As we continue to explore the potential of AI in web development and customer service, projects like the Alexela chatbot stand as a testament to our ability to blend technology and customer needs effectively. Since this is an ongoing project, we are adding new functions such as action triggers and more.
Stay tuned for more insights and innovations from the ADM Interactive team.