Role
Team
Collaboration
Duration
Highlights
Designed the first virtual AI building assistant that reduces discovery work effort by up to 90% and saves companies $1,500 by helping users access information quickly, change controls, and generate actionable insights.
Chat
ARIA Chat provides real-time answers to user queries about equipment, faults, and system status. It also enables users to perform automated tasks efficiently, and streamline troubleshooting.
Voice assistant
ARIA's voice assistant allows users to interact hands-free. Whether searching for information, adjusting settings, or accessing key features, users can stay productive without needing to touch their device.
A shrinking pool of skilled HVAC technicians, coupled with outdated training methods has left younger workers unprepared for complex diagnostics, slow troubleshooting and worsening industry-wide efficiency gaps.
-Anonymous Technician
Constraints
How might we help technicians to identify errors and troubleshoot HVAC systems effectively, even when they lack extensive experience or access to comprehensive resources?
Target Audience
Technicians: maintain and troubleshoot HVAC systems
Facility managers: oversee the operation and maintenance of building systems
Energy managers: monitor and optimize energy usage across buildings or portfolios
Design Process Overview
Competitive analysis
User research
Placement
Initial Design Concept
Refined Design
User feedback
Current Design
Research Insights
Competitors analyzed: 75F, Chat GPT, Co-Pilot
75 F
Focused on HVAC automation
Weakness: lacked advanced trouble shooting
Chat GPT
Set the design standards for GenAI chat interfaces
Weakness: generic use cases lacking industry-specific focus
Co-Pilot
Seamless integration with productivity tools and workflows
Weakness: limited application in specialized industries
HVAC
The HVAC industry currently faces a shortage of skilled technicians, compounded by a growing knowledge gap as experienced professionals retire. Younger technicians often lack access to comprehensive training, leaving them underprepared for complex, real-world challenges. This disconnect not only strains workforce efficiency but also risks the loss of invaluable expertise.
Technicians Painpoints
No real-time data: Relies on manual inspections
Inefficient documentation: Paper manuals or unsearchable digital files
Skill gaps: Junior technicians lack mentorship or access to institutional knowledge
Physical strain: Hazardous conditions (e.g., extreme weather, tight spaces)
Facility Manager Painpoints
Reactive workflows: Constant firefighting instead of preventive action
Communication gaps: Hard to translate technical issues for executives
Mutual Painpoint
Prioritization challenges: With multiple work orders flooding in, technicians lack clear guidance on which issues require immediate attention
Use Case
An inexperienced technician receives an urgent call: "Our office is sweltering—it’s 80°F in here!"
The client expects a quick fix, but the technician lacks prior knowledge of this building.
After a long drive in freezing weather, they arrive at an unfamiliar site.
Pain Point: No digital floor plans or equipment locations. They waste time searching for the rooftop HVAC unit, shivering in the cold.
Frustration: The technician flips through outdated paper manuals, struggling to match the system’s model to the documentation.
15 minutes pass just to identify the equipment, 15 to pinpoint the issue, and another 30 minutes of trial-and-error troubleshooting.
The adjustment works—but it’s a band-aid solution. No time for root-cause analysis.
As they pack up, another call comes.
Another hour-long drive and repeats the whole manual process.
Medium
Most users have tablets but some only have phones.
On the mobile app, users are limited to certain capabilities: equipment details, setpoint changes, and schedule changes.
User Goal
To reduce time spent on searching and prioritize their tasks.
To define ARIA’s optimal placement and footprint, grouped user needs by capabilities and evaluated spatial requirements: compact view (small space) vs modular overlay vs (medium space) vs expanded view (large space).
Information
Equipment details
Status
Work order
Space: medium
Automated tasks
Changing set point and schedules
Creating work orders
Space: medium
Detection
Alarms & faults
Equipment replacement
Predictive maintenance
Space: medium
Data analysis
Reports
KPIs
Graphs & tables
Space: large
Design Decision
Full-page feature with a chat and voice assistance interface to accommodate for the space needed.
After aligning with the CTO on key requirements, I distilled their vision into high-impact marketing assets designed to captivate investors. Understanding that visual appeal and intuitive storytelling was critical, I prioritized feature showcases through bold layouts —ensuring stakeholders could immediately grasp the product’s functionality.
In this iteration, the design prioritized user needs, while adapting to new constraints. During this phase, our company acquired another software platform, requiring a transition to their design. Simultaneously, as investment rounds progressed, the CEO emphasized enhancing key elements—such as the ARIA branding and voice assistant button.
Challenge
While the CEO championed a flashy ‘futuristic robot’ look, I partnered with the CTO to push back with alternative designs. By grounding our recommendations in accessibility standards and usability research, we advocates for a more simplistic, human-centered aesthetic.
The simple interface was inspired by many GenAI Chat such as ChatGPT, Deep Seek, and Perplexity.
The feedback is categorized it into two groups: technical issues and design/usability issues. Each problem is listed in the first column of the table and evaluated by its impact, feasibility, and priority from a design perspective, using a 3-point scale. Afterward, I collaborated with my PM to refine the priorities (yellow highlight) based on business needs.
Disclaimers
ARIA is in Beta
We need to catch up on design debt
We do not have a design system
Chat
A clean and intuitive chat interface designed for readability, with ample space for infographics and insights.
Voice Assistant
Voice Assistant enables hands-free conversations with a dynamic circular indicator that provides visual feedback. Users can easily tap to start or stop voice interactions, enhancing control and convenience. In addition, The real-time text display allow users to track the accuracy of their input.
Side Bar
Simply click the expand icon to view past chats. The "New Chat" button and conversation history is stored in the sidebar - organized by month for easy navigation.
Prompts, Export, and User Feedback
Guided Prompts – Starter suggestions appear at the beginning of each session to help users kickstart their conversations.
Seamless Sharing – Export any chat to easily share insights and key information.
Quick Feedback – Rate responses with thumbs up/down to help us improve user experience.
Current Ideation
User
Reducing discovery work effort by up to 90%
Saving up to $420 by eliminating the need for an initial site visit
Saving $1,500 by eliminating the need for repeat/recurring/new issue work orders
Company
Enhanced Customer Satisfaction – Simplifies HVAC management, improving user experience
Market Differentiation – Strengthens Brainbox’s position as a leader in AI-driven HVAC solutions
Increased Retention & Adoption – Provides value that encourages long-term customer engagement
Communication with Engineer
I worked with a developer who wasn’t particularly detail-oriented and didn’t prioritize design details. To address this, I took a proactive approach by consistently requesting updates, scheduling 1-on-1 meetings, and organizing weekly check-ins with a senior engineer who was passionate about UI changes. The senior engineer's involvement was crucial when the developer dismissed certain details or claimed that something wasn’t feasible. This experience taught me the importance of adapting my communication style, proactively reinforcing design standards, and involving additional stakeholders to ensure the designs were executed accurately.
Cross Functional Collaboration
Working on ARIA gave me the chance to collaborate with cross-functional teams: engineers, external agents, AI researchers, project tech leads, and sales. This taught me how to navigate different perspectives, balance technical feasibility with user needs, and ensure alignment across teams. By working closely with each group, I gained the context needed to advocate for design priorities while also considering technical constraints and business goals.
