OVERVIEW
COMPANY
CORE TEAM
TIMELINE
Jan 2024 - May 2024
TOOLS
ROLE
Lead Designer
RESPONSIBILITIES
Product redesign
Mobile design
Visual design
Interaction design
User research
Design system
Product strategy
ABOUT EQUINIX
Equinix is a global leader in digital infrastructure, operating 250+ data centers across 55 markets to deliver secure, high-performance colocation and interconnection services. By enabling direct connections between businesses, cloud providers, and networks, Equinix helps organizations accelerate digital transformation, improve efficiency, and scale globally.
PROBLEM
To stay ahead of the GenAI curve, Equinix’s Enterprise Architecture team—partnering with Sales, Operations Enablement, Human Resources, and others—developed several domain-specific GPT pilot applications. These tools tapped into internal knowledge to help employees summarize content, translate information, debug code, and optimize performance.
However, design was brought in only after functional prototypes were built. Without early UX involvement, the pilots lacked usability, visual consistency, and alignment with real user workflows—leading to rework, inefficiencies, and missed opportunities to deliver a seamless, intuitive experience.
APPROACH
As Design Lead, I defined a shared vision and cohesive design direction for a unified, company-wide GPT experience—Equinix-GPT (E-GPT).
The goal was to streamline multiple domain-specific tools into a single, intuitive application that elevates the employee experience across both mobile and desktop.
CHALLENGE
Multiple business silos—including Sales, Operations, and HR—owned their respective domain-specific GPT pilots, making it difficult to establish clear ownership of a unified, all-employee GPT solution. Competing priorities, fragmented decision-making, and the absence of centralized direction led to strategic misalignment and slowed progress toward a cohesive, enterprise-wide experience.
PROCESS
To align business units on a shared vision for E-GPT and collaboratively shape the foundational experience, Jung and I co-facilitated a series of Stakeholder Design Workshops using a human-centered design approach. These sessions brought together key representatives from Sales, HR, Operations, and other domains to break down silos, clarify ownership, and ensure the solution reflected cross-functional needs.
By placing users at the center of the process, we co-created the foundation for a cohesive, company-wide GenAI application—guided by the following key frameworks:
Jobs-to-be-Done
We used the Jobs-to-be-Done framework to uncover the core tasks employees aimed to accomplish with GenAI. This helped us align problem statements, prioritize use cases, and inform early product development based on actual user needs.
Killer Questions
To anticipate potential roadblocks, we applied the Killer Questions method—challenging assumptions and identifying risks that could derail implementation. These were explored through a hypothesis → design → prototype → test cycle, enabling proactive problem-solving early in the process.
II. DESIGN SYSTEM
To support the unique needs of an AI-powered, employee-facing application, Sal and I built a dedicated design system tailored specifically for E-GPT. This system addressed interaction patterns, components, and guidelines not covered by the existing enterprise library—ensuring design consistency, improving usability, and reducing development time.
Several critical factors made it clear that a custom design system was necessary to support the unique demands of the GPT application:
Unique User Interactions
E-GPT introduced novel workflows—such as natural language input, dynamic AI responses, and conversational flows—that required specialized components and patterns beyond those in the existing system.
Missing Core Components
Pre-existing UI kits lacked essential elements like chat interfaces, AI-suggested actions, and contextual tooltips, prompting us to design these from the ground up to maintain usability and cohesion.
Scalability & Flexibility
Given the evolving nature of GenAI use cases, the system needed to support rapid iteration and modular growth, with reusable components adaptable across domains and contexts.
Brand Alignment
As an all-employee tool, it was critical that the GPT experience felt unified with Equinix’s brand while accommodating the forward-looking nature of conversational AI.
III. IMPLEMENTATION MODEL
Together with Srivats, I defined a scalable architecture to enable a seamless, cross-domain GPT experience, grounded in the following key components:
Intent Recognition
Identifies the domain and context of a user’s query, prompting for clarification when needed to ensure accurate routing.
GPT Orchestration
Directs the query to the appropriate domain-specific API based on identified intent.
Advanced Orchestration
Supports multi-domain queries by extracting and consolidating data from various APIs into a unified response.
Guided Prompt
Offers users curated, domain-specific prompts to help narrow query scope and improve relevance.
Contextual Query
Allows users to continue conversations across domains, with contextual understanding carried forward to subsequent queries.
IV. FINAL DESIGN
The final design brings together key insights, user needs, and architectural decisions into a cohesive, intuitive experience—showcased below through high-fidelity wireframes.
V. SENTIMENT ANALYSIS
To evaluate the impact of E-GPT, I collaborated with Jung to conduct a sentiment analysis with five active beta users of the initial functional prototype. We used a mixed-method approach: the System Usability Scale (SUS) provided quantitative benchmarks before and after the design intervention, while open-ended interviews offered qualitative insights into Customer Satisfaction (CSAT), perceived value, and usage patterns.
Why Sentiment Analysis matters?
Goes beyond clicks and views to evaluate the quality of user interactions and overall experience.
Uncovers the “why” behind user behavior, surfacing emotional responses and personal perceptions.
Reveals how users perceive competitors, enabling smarter prioritization and differentiation.
Detects emerging trends and early shifts in sentiment—both positive and negative—for proactive product evolution.
IMPACT
LEARNING
Early Design Involvement is Critical
Late-stage design entry reinforced the importance of advocating for early UX involvement to reduce rework and align solutions with real user needs.
Cross-Functional Alignment Enables Scale
Navigating siloed ownership emphasized the value of a shared vision and unified goals when building enterprise-wide, scalable solutions.
Custom Tools Need Custom Systems
Designing for an AI-first experience required a tailored design system to support unique interaction patterns and ensure long-term scalability.
