Case Study

A voice-activated Digital Shop Assistant that transforms facilities inventory management

Digital Scientists partnered with HD Supply to design and build a tablet-based, voice-activated inventory management app and ordering interface for facilities maintenance teams -- delivered from concept to functional prototype in under 90 days.

Voice AI React Native iPad App Innovation R&D MVP

<90

Days from Concept to Prototype

4

User Personas Defined

3

Solution Concepts Explored

2

Input Modes (Voice + Touch)

Overview

From industrial distributor to integrated service provider

HD Supply, one of the largest industrial distributors in North America, faced a critical challenge from Amazon and online retail competitors. To stay competitive, the company's leadership determined it must transition from a traditional distributor and supplier role to that of an integrated service provider -- delivering technology-driven solutions directly to facilities maintenance teams.

As an innovation research partner, Digital Scientists ran a detailed ideation exercise envisioning the role of a property maintenance manager 5-10 years from now. Instead of focusing on the transaction of buying supplies, the team zoomed out to imagine augmented reality diagnostics, image recognition for part identification, and voice recognition for hands-free ordering.

Through this process, Digital Scientists identified the first-phase solutions that could be developed and launched with speed. Because technicians had limited access to technology, the best starting point was to make the storeroom smart -- building a voice-activated Digital Shop Assistant running on an iPad mounted in facility storerooms.

HD Supply logo

Client

HD Supply

Industry

Industrial Distribution, Facilities Maintenance

Services

Innovation R&D, Customer Journey Mapping, Product Design, Mobile App Development, MVP

Engagement

2017 (90-day sprint)

The Challenge

Manual inventory processes burdened maintenance teams across hotels and multifamily properties

HD Supply lacked an integrated system that would allow property managers to identify and manage assets, inventory, or tasks assigned to maintenance personnel. Facilities maintenance teams within hotel and multifamily industries were burdened by manual tasks -- spending excessive time searching for parts, making trips to hardware stores, and managing inventory by hand.

Part Lookup & Identification

Technicians spent too much time searching for specific parts and had no efficient way to identify items, locate replacements, or check bin locations.

Inventory Blind Spots

No system to predict inventory replenishments or alert when consumable goods were running low. Stockouts led to emergency hardware store trips.

Ordering Friction

No way to search for and buy parts on the spot. The ordering workflow was disconnected from the maintenance context, requiring multiple systems and approvals.

No Data Collection

HD Supply had no way to collect facility data on what inventory items were needed, where, and when -- missing critical insights to optimize their distribution business.

Amazon Competition

Amazon's growing industrial supply business threatened HD Supply's core market. The company needed to demonstrate innovation and deliver added value beyond traditional distribution.

Hands-Busy Technicians

Maintenance technicians often have their hands full with tools and parts. Traditional touch-only interfaces were impractical in the storeroom environment.

Jobs To Be Done

What maintenance teams needed to accomplish

Locate and identify specific parts quickly, with bin number and quantity information

Check in and check out inventory items linked to work orders and job numbers

Search and order replacement parts by voice, touch, text, or barcode scan without leaving the storeroom

Receive alerts when inventory is running low before a stockout disrupts maintenance work

Browse the HD Supply catalog and place orders directly from the storeroom kiosk

Track technician activity, inventory usage patterns, and order history to optimize procurement

User Research

Four personas spanning the maintenance ecosystem

Digital Scientists defined four key personas across the facilities maintenance value chain -- from the hands-on technician to the resident waiting for a repair. Each persona's goals and pain points informed the Digital Shop Assistant's feature priorities.

Maintenance Technician

Task-oriented, not tech-savvy

Goal: Locate, identify, and order parts faster

Pain: Spends too much time searching for specific parts and making trips to hardware stores. Needs to consolidate trips to reduce time and effort.

Maintenance Supervisor

Task-oriented, manages teams

Goal: Streamline purchasing workflows and track technician locations

Pain: Struggles to keep technicians moving and completing work orders in a timely and cost-effective manner.

Property Manager

Moderately tech-savvy, cost-conscious

Goal: Ensure customer satisfaction and control repair costs

Pain: Needs resident complaints handled quickly without repeat incidents. Unhappy residents reflect poorly on the community.

Resident

Tech-savvy, needs fast resolution

Goal: Know when technicians are coming and when issues will be resolved

Pain: Needs maintenance issues resolved quickly, especially critical items like leaks, electrical issues, and HVAC failures.

Our Approach

Three solution concepts evaluated, one fast-tracked to prototype

Digital Scientists explored three distinct solution concepts for the Digital Shop Assistant, each representing a different technology approach. The team evaluated pros, cons, and workflows for each before selecting the Digital Assistant Appliance as the first-phase solution to prototype.

Digital Assistant Appliance

Selected for Phase 1

A tablet-mounted hub in the storeroom supporting voice search, touch-screen display, and barcode scanning -- without requiring technicians to carry a device.

Pros: Cost-efficient, limited hardware required, easier learning curve since scanning activates current jobs and information.

Wearable Device

Future Concept

Google Glass or similar AR wearables providing heads-up display for real-time product inventory and voice-driven part identification.

Pros: Hands-free operation, real-time information, future AR capabilities for field diagnostics.

Vision-Based System

Future Concept

Automated inventory control using AI-visual observation and shelf sensors -- similar to Amazon Go -- for seamless user and part identification.

Pros: Fully automated identification, preserves natural shopping behavior while offloading data tasks to automation.

Input Design

Three input modes designed for novice to expert users

The Digital Shop Assistant was designed to support multiple interaction paradigms, allowing technicians to transition from traditional touch mode to voice-only as they gain experience. Each mode addresses different use case complexity levels.

01

View Screen + Voice

Screen guides the user along. Can stand a few feet away from the device. Works well if the user does not have commands memorized.

02

Voice Only

Commands and actions need to be memorized. Does not need to look at the device. Works well for small, routine interactions.

03

Touch + Voice Hybrid

Some actions are easier with direct screen interaction. Needs to stand close to the tablet. Works well for complex interactions like catalog browsing.

Product Design

Voice-activated inventory management with the “Hey FM” hotword

The Digital Shop Assistant responds to the hotword “Hey FM” followed by natural voice commands. Common actions are listed on the dashboard, search results display with real-time inventory levels, and contextual voice help is always available -- just ask “Hey FM: What can I do?”

Voice search results with inventory status indicators

Product detail with pricing, location, and quick actions

Hotword Detection

Snowboy hotword detection activates the assistant when technicians say “Hey FM.” Contextual voice commands are displayed on screen for guidance.

Real-Time Inventory

Search results show pricing, bin location, quantity remaining, and stock status. Running Low and Out of Stock indicators alert technicians immediately.

Voice + Touch Actions

Check in, check out, add to cart, search, browse catalog -- all available through voice commands or touch interaction, adapting to user preference.

Voice Commands

Simple and compound voice commands for every inventory task

Digital Scientists designed a comprehensive voice command system supporting both simple one-word commands and compound natural-language phrases. The system was built to handle the challenges of voice control -- including multilingual support, noisy environments, and rigid syntax -- with contextual on-screen guidance.

Simple Commands

“Hi FM”

Wake up assistant

“Search inventory”

Search on-site items

“Check in”

Return items

“Check out”

Take items for job

“Help”

Show available commands

“Mayday”

Tech support call

Compound Commands

“We're running low on [item name]”

Prompts for quantity and location details

“What's the price of [item name]?”

Looks up item and provides pricing

“Where are the air filters located?”

Provides bin location for item

Use Case Scenario

Simple part checkout: from walk-in to walk-out in under a minute

Digital Scientists designed the primary use case around the maintenance technician's storeroom visit. The scenario demonstrates how the Digital Shop Assistant transforms a multi-step manual process into a streamlined voice-driven interaction.

Conversational Flow

1

Identification

Joe: “Hey, this is Joe.” → DA: “Hi Joe. Showing your dashboard.”

2

Part Search

Joe: “I need a faucet.” → DA: “Showing inventory for Faucets.”

3

Selection & Checkout

Joe: “I need the bath faucet from Unit 201. Check out this item.” → DA: “Okay. Item checked out.”

4

Low Inventory Alert

DA: “We're running low on these items.” Joe: “Let's get 5 more of the first one.” → DA: “Adding items to cart.”

5

Exit & Summary

Joe: “Okay, that's all.” → DA: “Logging you out and sending a summary to your phone.”

Primary Use Case Flow

Tech walks in and is recognized via NFC, Bluetooth, camera, voice, or ID badge scan

Selects a job number -- parts required for that job are shown automatically

System shows item location, bin number, and quantity remaining

Inventory system records part checkout and updates counts

Out-of-stock items can be added to cart for ordering via voice or touch

Push notifications sent to technician and maintenance manager with summary

Work order, cart, tech location, and time are all updated in real time

Technology Stack

Built with React Native and Snowboy hotword detection

Digital Scientists selected a technology stack optimized for cross-platform deployment and voice interaction. The team also conducted extensive comparative research across voice platforms, hardware options, and interaction patterns to inform the design.

Core Technology

React Native -- cross-platform mobile framework

Snowboy Hotword Detection -- on-device “Hey FM” wake word

Native iOS & Android -- dual platform deployment

Swagger API -- standardized API documentation

SSML -- Speech Synthesis Markup Language for voice responses

Comparative Research

Google Assistant & Lens -- voice and image recognition patterns

Apple Siri -- on-screen help state patterns

Amazon Alexa & Dash -- voice skill design and IoT ordering

Amazon Go -- vision-based automated checkout

Square POS & Houndify -- kiosk hardware and NLP APIs

Results

A functional prototype that signaled innovation to HD Supply's customers

In just over 90 days, Digital Scientists built a functional prototype that demonstrated HD Supply's commitment to innovation and value creation. The prototype opened lines of communication with customers, alerting them that technology-driven efficiencies were on the way.

While other companies begin at the wrong point and build an app without talking to customers first, the HD Supply team acknowledged the importance of focusing on the customer first -- starting the process with a deep understanding of maintenance teams' needs through persona research, comparative analysis, and use case scenarios.

90

Days from concept to working prototype

4

Personas researched across the maintenance ecosystem

13+

Voice commands designed for simple and compound interactions

3

Input modes: voice, touch, and hybrid

Benefits Delivered

More efficient inventory management system with real-time tracking

Increased customer and guest satisfaction through faster repairs

Better inventory planning with predictive low-stock alerts

Detailed customer data on where and what inventory items are needed

Deliverables

Persona and stakeholder assessment

Comparative technology research (voice AI, hardware, IoT)

Use case scenarios and voice command design

Wireframes, visual design, and clickable prototype

Functional React Native prototype with voice integration

Next Case Study Holbrook Life →

Ready to innovate with voice AI and IoT?

Digital Scientists designs and builds voice-activated interfaces, IoT solutions, and enterprise mobile apps. Let's discuss how we can bring innovation to your operations.