Wendy's X Google Cloud


Position:      
Creative Technologist Intern at Deeplocal
Role:            
Designing and building a logging system to facilitate easier debugging and benchmarking for usability tests.
Team Lead:
Sean Scanlan

Client:        
Google Could Next’23 AI Conference



As a Creative Technologist Intern at Deeplocal, I contributed to the development of interactive exhibitions for the Google Cloud Next 2023 conference. One notable project was the “Wendy’s x Google Cloud” exhibit, which showcased Wendy’s Fresh AI, a conversational AI platform utilizing voice recognition technology powered by Google’s Vertex AI.

In this project, I was responsible for designing and building a logging system to facilitate easier debugging and benchmarking for usability tests.

This role provided valuable insights into industry best practices and the integration of AI models. The rise of AI assisted voice user interfaces (VUI) and their incorporations in the existing applications is an exciting field to look into further.















The Problem:

    The challenge was to streamline the usability testing process for Wendy’s Fresh AI, which required handling large volumes of data generated by the AI’s interactions with users. The existing system needed enhancements to efficiently log and interpret these interactions to identify areas for improvement, to insure smooth functioning of the exhibit at the event.




    Key Features:



    • Integration with Fresh AI: The logging system was directly connected to Wendy’s Fresh AI LLM model. Each time a user placed an order, the system communicated with the LLM model and received responses formatted as JSON outputs.

    • Asynchronous Processing: Designed to handle large batches of orders asynchronously, ensuring the system could process data in real-time without delays.

    • Human-Readable Outputs: Converted the machine-readable JSON data into human-readable formats, making it easier for the team to analyze and interpret the results.

    • Usability Testing Enhancement: Facilitated more efficient and effective usability testing, allowing the team to quickly identify and address any issues with the AI’s performance.