researchers at honda spend hours creating new product proposals. our research exploration finds new ways to incorporate generative ai in these workflows.
Timeline
Jan 2024 - May 2024
Team
1 Designer
1 PM
2 Engineers
CONTEXT
honda researchers scrape through mobility datasets to find opportunities for innovation.
Honda Research Institute (HRI) focuses on exploring opportunities in which Honda could be early innovators in the mobility space. Having seen the impact and potential of generative AI, HRI enlisted our team to create an MVP for an internal tool that explores new uses for generative AI within their product innovation teams
“We just want to see what LLMs can realistically do for us and how far we can take that.” - Ryan Lingo
APPROACH
go through discovery to suggest gen ai interaction patterns that could be implemented at honda
PINPOINTING PROJECT SCOPE
research and product teams are encouraged to take creative risks
At HRI, product innovation begins with data. Research teams focus on compiling datasets, analyzing clusters, and providing findings to the product team. Based on our conversations with the HRI team, we narrowed down the project to cater to the following 2 personas.
insight 1: generative ai can expedite processes and widen the scope of exploration to arrive at more innovative and groundbreaking discoveries
IDENTIFYING USES
different levels of human-ai teaming are needed at each stage
We found that while innovation and creative risks are encouraged, they can often be held back by time constraints that limit the breadth of exploration. This set up the groundwork to work with the team to identify steps in which the MVP could both alleviate pain points and push innovation.
insight 2: generative ai should not replace all functions but rather serve as a collaborative aide that requires different levels of supervision at each stage
CONCEPT IDEATION
llm outputs beyond a chatbot
We wanted to explore other methods of interaction between the AI and the user. We went through several rounds of concept ideation to think of new methods of interaction and outputs.
insight 3: limiting inputs, using explainable ai, and custom interfaces are alternative ways to ensure clear communication and reduce misinformation
PRIORITIZATION FOR DEVELOPMENT
high autonomy, low restriction
We evaluated our concepts on a matrix of user autonomy vs. LLM restriction level to prioritize building out the concepts that had high autonomy and low restriction. Based on this framework, we identified features that could be explored now, features that could be a bonus if explored, and features that were not right to explore at this time.
insight 3: in order to maximize impact, the tool should provide high user autonomy to encourage creative thinking, while also restricting outputs through standard and prompt engineered formats to reduce hallucination risk and ensure accuracy
see ya next time :)
last updated august 2025
exploring gen ai use cases
honda research 2024