Yardflow VDA – United Rentals

AI Voice Driven Application

United Rentals partnered with us to build a hands-free, AI-powered voice assistant for Equipment Associates (EAs) to streamline yard operations, specifically equipment returns and sorting.

Technologies

Business Challenge

Equipment Associates at United Rentals faced inefficiencies in daily yard operations. Operators had to pause inspections to manually enter data, often relying on paper notes before updating the system. Searching for equipment was also slow, requiring manual ID checks because existing apps lacked visual identifiers like brand or color.

What We Did

Voice-to-Voice (V2V) Interaction

Built a natural, human-sounding conversational interface that allows EAs to execute workflows entirely through voice, maintaining focus on the equipment rather than a screen.

Agentic AI Architecture

Implemented a Supervisor-Worker hierarchy where a Supervisor Agent manages session context and delegates specialized tasks to dedicated sub-agents for Equipment Return and Equipment Sort.

The Complexity Mitigation

o Digital "Draft" State: The AI acts as the "paper buffer" by recapping verbal inputs for confirmation before final submission.
o Visual AI Pipeline: Developed an asynchronous image processing pipeline that captures photos (via phone) to identify equipment and document damage without interrupting the voice flow.

Voice-to-Voice (V2V) Interaction

Built a natural, human-sounding conversational interface that allows EAs to execute workflows entirely through voice, maintaining focus on the equipment rather than a screen.

Agentic AI Architecture

Implemented a Supervisor-Worker hierarchy where a Supervisor Agent manages session context and delegates specialized tasks to dedicated sub-agents for Equipment Return and Equipment Sort.

The Complexity Mitigation

How We Did It (The Implementation)

Architectural Excellence

Built a robust backend to handle the low-latency V2V communication and agentic delegation.

AI Tech Stack

Leveraged our internal AI tools (MaxQ) to build, monitor and fine tune the AI infrastructure behind the app, allowing for fast iterative cycles and a UX optimized to the users' needs.

Hardware Ecosystem

Seamlessly integrated with Samsung Galaxy S25 (Android >14) and commercial Bluetooth headsets.

Observability

Integrated Sentry for high-fidelity audio logging and session replay to diagnose environmental noise and hardware performance in the field.

Key Features Delivered

Functional Android Application (APK)

Optimized for background processing, enabling uninterrupted voice chats and image uploads even when the phone screen is locked.

AI-Powered Equipment Agents

Delivered intelligent Equipment Return and Equipment Sort agents capable of collecting equipment-specific data through dynamic conversations and enabling faster routing decisions with concise, task-focused summaries.

Translation Layer for Legacy Systems

Built a translation layer that converts complex, code-heavy API responses from the UR legacy backend into natural language, significantly improving AI understanding and accuracy.

Functional Android Application (APK)

Optimized for background processing, enabling uninterrupted voice chats and image uploads even when the phone screen is locked.

AI-Powered Equipment Agents

Translation Layer for Legacy Systems

Built a translation layer that converts complex, code-heavy API responses from the UR legacy backend into natural language, significantly improving AI understanding and accuracy.

What was the result

Polar Stork brought a strong mix of product, UX, and technical expertise to the project. Their UX mapping was exceptional and helped uncover important edge cases while refining the overall experience. The team quickly demonstrated a deep understanding of our workflows, and their transparency and collaborative approach gave us confidence throughout the engagement.