Brendan Appe

AI-Augmented Inventory Search & Build
We explored a generative interface for Toyota’s customer-facing vehicle inventory systems. It helped users assemble custom vehicle packages and search availability across dealerships using conversational and criteria-driven prompts.

AI integrated seamlessly into the discovery experience, allowing users to describe their needs naturally and receive guided, personalized support.

Recommendations reflect user context, making it easier to compare and customize vehicles suited for specific lifestyles.

To ensure the system behaved predictably and intuitively, I led working sessions with engineering and business teams to define how the AI should operate across three key phases of the journey:

Define intent
Build vehicle
Search inventory.

For each phase, we mapped inputs, outputs, adaptation logic, and failure states to define the appropriate patterns.

Sample of our input, output, and adaptation logic framework.

Fleshed-out example that walks through a user interaction across each layer.

We ran common customer scenarios through each layer of the system to document how the AI should respond in real-world contexts. This helped us capture edge cases, ensure consistency, and guide both design and technical implementation.

Before we could complete this exercise effectively, we first needed to align on tone of voice. I facilitated personality-mapping sessions for each AI application to define how the system should sound and behave.

For customer-facing experiences, we landed on The Guide: a warm, calm, and confident archetype focused on empowerment and support.

This alignment helped translate design intent into system behavior, ensuring a more intuitive and resilient experience.

During vehicle build, the assistant adapts in real time, prompting clarifying questions based on user input.

As preferences are selected, the system updates chips and trim recommendations in real time to reflect the customer’s needs.

Users can opt-out of AI assistance at any point, returning to the standard vehicle build experience.

Turning off AI assistance removes tailored guidance—returning users to a broader, unfiltered build experience without personalized recommendations.

At the end of the journey, AI surfaces vehicle matches based on customer preferences and real-time supply chain availability.

AI Assistant for Supply Chain Teams
Integrated within Toyota’s Cube Supply Chain Management Platform, we designed an AI interface enabling team members to query and interact with logistics data using natural language prompts.

The AI interprets user input and surfaces key patterns, offering contextual next steps.

For internal Toyota team member experiences, we chose The Sage: a composed, thoughtful, and trustworthy voice optimized for clarity and decision-making.

The interface dynamically shaped itself based on the user’s request. This reduced time spent navigating complex dashboards and helped teams surface the right insights faster.

Additionally, I led design efforts to identify and implement more tactical applications of AI within Toyota’s supply chain platform. One of the most effective areas of application was making complex rule-based automations more transparent and readable for business users.

Working in close collaboration with IBM’s watsonx.ai team, we integrated IBM’s Granite large language models to generate natural language summaries for multi-condition logic involving event and vehicle triggers. These summaries allowed non-technical users to quickly understand what would happen when certain thresholds were met.

This layer allowed business teams to confidently create, validate, and communicate logic without relying on technical support.

Users define multi-condition logic using event and vehicle-level conditions to automate platform behavior.

Upon collapse, natural language summaries clarify when rule conditions are met and actions will trigger.

These efforts remain active works in progress under NDA. Demos and deeper walkthroughs are available upon request.