Driving AI Design

The goal was to research vehicle forensics, a new area for us, and then design a prototype for an application that can be used to unobtrusively collect vehicle data via the infotainment USB and the vehicle's OBD II port. We conducted dozens of interviews with law enforcement agencies in North America and the UK to understand how they conduct vehicle forensics today and what our product would need to do to be a useful resource to them.

Key activities:

  • Interviewing customer users who performed vehicle forensics to understand their tools and processes.
  • Reviewing competitor offerings and existing workflows
  • Mapping out the application flow from lookup, to connection, through to acquisition of data.
  • Building an interactive prototype to test with users.

Researching vehicle forensics

Through extensive interviews, we spoke to many practitioners of vehicle forensics. They covered a wide range of experience levels and technical competency and available resources. Through the knowledge gained through this research we developed an understanding of the different personas, which helped when we designed the application.

Reviewing competitor offerings

In tandem with interviewing users, we reviewed the primary competitor offering to see how our product fit within the ecosystem. As they offered a mostly hardware-based solution, our unobtrusive software solution filled a void for customers limited by knowledge, time, and expense.

Mapping out the application

Before starting the application, I worked with the principle developer to understand the application workflow. Since the product was interacting with a wide range of vehicles, we needed to account for a variety of exception conditions that could occur during key parts of the operation. Through this exercise, we solidified an approach to guiding the user through the connection and acquisition process.

Building the prototype

The application consisted of two main parts:

  • The primary part for connecting to the vehicle and acquiring its data.
  • A lookup section for identifying vehicles and infotainment support.

    • The design for the primary part of the application was initially completed in Figma and then imported into Claude Code via the Figma MCP connector. From there, I iterated through several designs based on feedback from the team, internal stakeholders, and customer users.

    Primary flow: vehicle data acquisition

    To design the lookup section, the product manager and I wanted to explore a few different designs using AI tools so that we could share different approaches with some prospective users. The company was experimenting with Lovable as a design tool so we designed three different versions. Due to model costs, we decided to export the code and continue develop with Claude Code. I rapidly built out the three versions and we held design exploration sessions with customers. Of the three designs, one was a clear winner for reasons including: familiarity with layout, discoverability of UI affordances, and general aesthetics.

    Vehicle lookup flow: The winning design

    Results

    From the extensive research interviews of our customers, rapid prototying using AI, and several iterations of testing, we produced a functioning prototype that served as the basis for the product.