Cloud Theory AI

Cloud infrastructure teams were overwhelmed by data but starved for insight. Dashboards presented raw metrics without context — requiring deep expertise to interpret and slowing down decision-making across the organization.

Teams needed a way to quickly understand the state of their infrastructure at a glance, without sifting through dozens of charts and alerts. The reporting experience was fragmented, and onboarding new team members to the existing tooling took far too long.

Cloud Theory needed a platform that could transform dense infrastructure data into AI-generated summaries — delivering actionable intelligence instead of raw numbers, and making every team member immediately productive.

I conducted user research and usability testing to validate design decisions and guide product direction. This included interviews with infrastructure engineers and DevOps teams, analysis of existing reporting workflows, and competitive audits of dashboard tooling across the market.

Three critical patterns emerged from our research that reshaped our entire approach to the product.

Our research revealed that the core problem wasn't a lack of data — it was a lack of meaning. Teams had plenty of metrics but no way to quickly understand what they meant for their operations.

We reframed our design challenge:

This reframe shifted our focus from building more dashboards to designing an AI-powered reporting experience that tells the story of your infrastructure — meeting users with insight instead of data. From there, three guiding questions framed every design decision that followed:

I explored multiple dashboard concepts — each taking a different approach to surfacing AI-generated summaries. Leveraging AI tools including Subframe, Cursor, Figma Make, Claude, and ChatGPT, I was able to rapidly prototype and iterate at a pace that would have been impossible with traditional workflows.

This AI-augmented process allowed us to test structural assumptions quickly before investing in high-fidelity design, running concept tests with users to validate information architecture and interaction patterns.

The final design synthesized the strongest elements from our explorations into a dashboard experience centered around AI-generated summaries — giving users faster at-a-glance reporting that transforms raw infrastructure data into plain-language insights.

I also improved the design system's efficiency to accelerate adoption across both design and engineering teams, ensuring consistent implementation and reducing handoff friction throughout the product.

Special attention went into how AI summaries are presented — using progressive disclosure so users can start with a high-level overview and drill into details on demand. The transition between summary and detailed views needed to feel seamless and maintain spatial context.

I designed loading states that give users confidence the AI is processing their data, error handling that gracefully degrades to manual views, and empty states that guide new users toward their first meaningful interaction.

One of the most impactful decisions was refreshing the core visual language. The legacy dashboard relied on hard corners, flat contrast, and a dated color palette that felt disconnected from the modern tools our users open every day. We updated the system to feel contemporary, trustworthy, and at home alongside the design tools teams already rely on.

Rounded corners — We moved from sharp 2px corners to a softer, more modern radius scale. Rounded geometry creates a friendlier read at-a-glance and matches the visual conventions of modern SaaS tools, reducing the cognitive load of scanning dense data.

Higher contrast — We pushed text, surface, and accent tokens toward higher contrast ratios to improve scannability and meet accessibility standards. This was especially important for the dense inventory tables and AI summary panels, where users need to parse information in seconds.

Modern dark-mode palette — We rebuilt the palette to feel native to today's design tools — closer in tone to Figma, Linear, and Vercel — using cooler neutrals and a more saturated brand accent. This made the product feel current, signaled product maturity, and gave our design system colors that teams across the organization were immediately comfortable adopting.

I conducted usability testing to validate design decisions and guide product direction. Through iterative testing rounds, we refined the AI summary format, the dashboard layout, and the design system components to ensure they met real user needs.

Key findings from testing directly shaped the final product — from how summaries are structured to how users navigate between overview and detail states. Each iteration was informed by direct user feedback and behavioral observation.