OVERVIEW

Problem: In manufacturing facilities, defense logistics networks, and distributed enterprise environments, non-technical operators manage thousands of edge devices manually and reactively. They lack a centralized control, and a reliable way to detect issues before they become operational crises. They needed something fundamentally different. An experience that could see everything, act before things broke, and guide them without requiring them to become experts in what was running underneath.

Solution: An agent-led edge management platform, an experience that gives technical and non-technical operators one intelligent place to onboard devices, monitor fleet health, manage applications, and catch vulnerabilities earlier to prevent crises entirely, while introducing an agentic layer that runs continuously beneath the surface. It checks every connected device every sixty seconds, detects vulnerabilities, triggers automated recovery, and restores affected devices to their last known secure state without waiting for human intervention. The system does not react to crises. It prevents them.

My Role & Impact: I joined a 15-person engineering and architecture team and immediately introduced two significant changes.

1. I established a research-led UX process that shifted product decisions from technical assumptions to validated customer needs.
2. I introduced an AI-augmented design delivery model that moved the team from static handoffs to working, implementation-ready experiences.

These changes created the foundation to define the core experience, unify Edge, Ansible, and OpenShift through a portable information architecture, and introduce an agentic layer that could understand each customer’s environment and guide them toward the next relevant action. Within 6 months, my team and I established a market-ready foundation for a $10M platform.

• Redirected a 15-person engineering and architecture team toward customer-validated priorities.
• Unified 3 products through a portable information architecture for operators managing 10,000+ connected devices.
• Established an agentic experience that made complex environments more accessible to non-technical operators.
• Improved cross-functional delivery efficiency by 30% through AI-augmented design and implementation-ready prototypes.

UX PROCESS

DISCOVERY

The team consisted of 15 engineers and architects with deep technical expertise, but no clear product roadmap, customer research, or shared vision for General Availability, Feb, 2026. The product was being shaped primarily around assumptions and existing technical capabilities, creating a significant risk: we could build something technically impressive that customers did not understand, value, or adopt.

I recognized a critical need to bring real customer evidence into the room. Through the Early Access Program, I recruited 10 enterprise clients across manufacturing, defense logistics, and distributed infrastructure, giving the team something it had never had before: direct insight into the people who would buy and use the product. I led research sessions and interviews with enterprise partners, including Eli Lilly, ABB, Cox Automotive, and Lockheed Martin. What we heard reframed the entire initiative. Two distinct groups were shaping the product’s success, and each needed something different.

• Buyers were focused on business outcomes: reducing operational costs, strengthening security visibility, maintaining consistent performance across distributed sites, and scaling without adding headcount.
• Users needed clarity and control: a fast path to connect a device, an understandable view of fleet health, and the ability to resolve a problem without needing to understand the infrastructure operating beneath the experience.

Until that point, these needs had never been brought together within a single product strategy. The research gave us the foundation to connect business value with a usable, trusted experience, and begin shaping a product that could succeed with both buyers and users.

That finding changed the strategic direction of the product. It told the team exactly where to put our energy for the first release and why. Earn trust with the core experience first. Then introduce intelligence on top of it. The 10% of customers open to AI and automation were not rejecting it. They were telling us the order of operations. Foundation before intelligence. Reliability before automation.

From those sessions I built persona frameworks spanning four buyer stakeholder groups and three core operator roles: Admin and System Integrator, Operator, and Viewer. For the first time, product, engineering, and design had a shared evidence-based picture of who they were building for and what each person needed to succeed.

Buyer personas 

User personas 

Through close partnership with product and engineering teams, I translated research insights into product themes, JTBDs, prioritized capabilities, and phased experience recommendations. This helped the team understand what needed to launch first for GA 1.0, what could evolve next, and how each release could move us closer to the desired future-state experience.

At enterprise scale, teams often move quickly within their own product areas, and over time it can become difficult to maintain visibility across shared customers, timelines, and dependencies. That was beginning to happen across Edge, Ansible, and OpenShift, where each team was advancing important work, but with limited visibility into how those efforts connected across the broader ecosystem.

I looked inward to better understand those connections. I evaluated existing workflows, intake processes, risks, and product surfaces to identify where teams could align earlier and share more context. The findings reinforced what we had already heard from customers: the products were part of the same experience, even though they were being developed through different paths. I introduced a more connected way of working that helped teams share context sooner, understand cross-product dependencies, and make decisions with the broader customer journey in mind.

High-level 

Define

The process was deliberately co-creative. Rather than presenting a finished designs for sign-off, I brought cross-functional teams into the work early, running sessions where we mapped what customers needed to accomplish, where existing flows broke down, and what a seamless experience could look like if we designed around the operator first.

Every step of that end-to-end journey was validated with the 10 enterprise clients in our Early Access Program before it was built. Customers reacted to the flow, identified where they felt uncertain, and confirmed where the experience finally matched how they actually worked. Their input shaped the final direction and gave engineering a journey that had real evidence behind every decision.

User journey – phase 1

This strategic approach helped me shift the experience from a feature-led path toward a product-led growth model. The goal was to help internal teams think customer-first by enabling customers to discover value within the environments they already used before asking them to adopt more broadly or expand.

To build alignment, I translated the complexity into an ideal-state customer journey and partnered with product and data teams to define signals for adoption, self-service, and upgrade readiness. This created a clearer path from discovery to adoption and expansion, allowing customers to start small, realize value early, and grow naturally while giving the business a more scalable approach to product-led growth.

User journey – phase 2 – Agentic experience 

The more consequential work extended across the design organization. I brought together the designers responsible for Edge, Ansible, and OpenShift to evaluate the three product experiences as one connected ecosystem, something the teams had not done before. Together, we reviewed navigation patterns, components, terminology, and interaction models to understand where the experiences aligned and where they diverged.

From that work, I established a unified information architecture, portable navigation system, and shared component foundation that could extend across Edge, Ansible, and OpenShift without disrupting the experience as users moved between products. I developed the approach collaboratively through working sessions, shared decision-making, and continuous input from each design team. The robust information architecture became the backbone of the broader platform, allowing capabilities to be introduced incrementally as each release validated new customer value.

Information architecture alignment 

IDEATE

I encouraged my team to start with key experience touchpoints, and  use AI as a way to accelerate alignment, make ideas visible earlier, and move with more confidence. I role-modeled this by implementing AI-augmented workflows myself first. I used Figma Make to generate low-fidelity wireframes, compare design directions, clarify requirements, and translate abstract product goals into tangible experience options. This helped the team see how AI could support better thinking, not replace design judgment.

e2e user flow (high-fidelity, iteration 2 – light mode)

Overview – Entry point (AI-augmented, high-fidelity, iteration 2 – light mode)

Image builds (AI-augmented, high-fidelity, iteration 3 – dark mode)

 

EXECUTE

The second thing I changed was how the team moved from ideas to alignment.

I introduced AI-augmented prototyping as the primary hand-off tool. The goal was not to automate design. The goal was to collapse the distance between an idea and a conversation about it.

This approach created a clearer path from idea to product sign-off to development readiness. It improved visibility for product and engineering partners, helped designers stay focused on the right level of fidelity at the right time, and contributed to an approximately 30% improvement in cross-functional delivery efficiency.

The full AI-enabled prototype was made available in GitLab here., which allowed engineers to review the working direction directly, contribute through pull requests, and refine the prototype alongside the design team.

Impact

Customers who once managed edge devices through emergency calls and manual intervention now onboard fleets in minutes, monitor thousands of devices in one place, and trust that the system is acting before they need to. The reactive scramble that defined their day became the exception rather than the norm.

For Red Hat, the work introduced a new workstream entirely. A unified platform architecture that could grow with customers, drive product-led expansion, and create the conditions for deeper AI adoption across the enterprise. The agentic layer was not just a feature. It became the proof of concept for what an intelligent Red Hat experience could look like at scale.

That direction did not exist in isolation. IBM and Red Hat’s combined commitment, backed by a $5 billion investment in AI and agentic capabilities, created the strategic environment that made this work possible and gave it a larger stage to land on. The platform we built was not ahead of its time. It was exactly on time, aligned to where the business was already moving and ready to grow alongside it.

• 30% improvement in cross-functional delivery efficiency tied to AI-augmented prototyping, clearer design standards, and a delivery model that reduced misalignment before it became expensive
• 10 enterprise clients validated the product direction through the early access program before a single line of production code was written
• 90% of validated customer needs centered on core capability, keeping GA 1.0 focused, shippable, and meaningful from day one
• A phased roadmap through phase 5, with the agentic upsell and capability recommendation layer sequenced to activate after trust was earned with the core experience

Integrated across OpenShift, RHEL, and Ansible (AI-augmented, light mode)