CASE STUDY
Google Cloud Platform
Simplifying the Cloud with Active Assist
The GCP needed automated recommendations to help users navigate the dense complexity of cloud management.
Project Challenge
Often navigating the cloud’s immense digital landscape is filled with complex calculations, technical configurations, and network puzzles that demand constant attention.
Cloud technology empowers incredible possibilities, but it can also bog users down. Google brought me on to help flip this challenge on its head. The challenge? To harness Google Cloud’s data, machine learning, and artificial intelligence to make cloud management intuitive for over five million users worldwide.
The Mission
Our goal was ambitious yet straightforward: to reduce cloud complexity by delivering proactive, user-centered recommendations across Google Cloud’s ecosystem.
By creating a single hub for personalized insights—spanning everything from billing to security—we aimed to empower users to make informed decisions quickly and confidently, minimizing the steep learning curve of cloud management.
My Role and Approach
I joined a two-designer team where I focused on crafting user flows, interface, and visual design for Active Assist’s recommendation hub.
My approach involved hands-on collaboration with engineers, product managers, and researchers to ensure our solutions were well-informed and cohesive. Over the year, I coordinated with teams across billing, security, analytics, and storage, developing an understanding of both the technical aspects of our product and the day-to-day needs of our users.
Learning the Landscape
To set the foundation for an impactful design, I invested time early on to understand the team’s collective knowledge and the users’ pain points.
Through coffee meetings and exploratory conversations, I gained insight into recurring issues: users were spending an inordinate amount of time on network troubleshooting, over-provisioning resources, and managing overly broad permissions. These discoveries helped us align on a shared mission: developing tools that would allow users to manage their systems without technical overwhelm.
Personas
To design effective cloud recommendations, we focused on understanding the unique needs and challenges of our primary users—IT Leaders, Engineers, and Data Scientists—each with distinct goals, pain points, and workflows within the cloud environment.
IT Leaders
Goals
-
Ensure the cloud environment is secure, compliant, and efficiently managed.
-
Optimize costs by eliminating unnecessary resources and improving resource utilization.
Pain Points
-
Overwhelmed by time-consuming permissions management and troubleshooting tasks.
-
Limited visibility into resource utilization, leading to potential inefficiencies.
Needs
-
Security and Compliance recommendations to manage permissions and enforce best practices.
-
Cost Optimization suggestions that show clear outcomes, helping them confidently implement changes without risk to security or functionality.
Engineers
Goals
-
Maintain system uptime and performance, minimizing latency or interruptions.
-
Implement configurations that align with best practices and scale with system growth.
Pain Points
-
Constant troubleshooting and complex configuration requirements, which can be time-consuming and prone to error.
-
Challenges balancing system performance with resource allocation and cost constraints.
Needs
-
Performance Optimization tips tailored to the system’s operational requirements.
-
Predictive Outcomes to see the impact of configuration changes in real time, ensuring that any adjustments enhance performance without increasing resource costs unnecessarily.
Data Scientists
Goals
-
Maximize data processing efficiency, enabling quick, reliable insights.
-
Access the necessary resources for data analysis without delays or permissions barriers.
Pain Points
-
Frustrations with resource allocation and delays due to limited access or over-restricted permissions.
-
Complexity in managing large data sets and data flows in real time, especially when resources are shared.
Needs
-
Resource Allocation suggestions to ensure they have access to adequate processing power for data tasks.
-
Permissions Management recommendations that enable seamless access to necessary resources while maintaining data security.
Research and Key Design Principles
To create a recommendation system that users could trust, I analyzed industry leaders like Terraform and AWS, studying their approaches, as well as user feedback, and identifying opportunities to make recommendations more actionable and intuitive.
From this research, I identified three key principles:
Recommendations in Context
Place recommendations where users need them—such as billing insights directly on the billing page—to reduce friction and streamline decision-making.
Show Predictive Outcomes
Complex recommendations, such as permission adjustments or billing changes, should show users the potential impact of their choices, helping them act with clarity and confidence.
Provide Reversible Options
Recognizing that cloud configurations are high-stakes, we designed an “undo” option wherever possible to allow users to experiment without irreversible consequences.
Rapid Prototyping and User-Centered Iteration
In partnership with our dedicated UX researcher, I led user testing sessions that explored how different user types interacted with recommendations. Our iterative prototyping process allowed us to quickly incorporate user feedback and refine the design.
By observing live tests with engineers, IT managers, and data scientists, I gained direct insights into the ways our recommendations impacted their workflows, helping us align the design with real-world needs.
These principles informed each design iteration, ensuring the hub would not only provide valuable insights but also encourage action by meeting users where they were.
THE SOLUTION
Active Assist’s Recommendation Hub
The final product was a powerful yet intuitive recommendation hub with three main components:
Overview Hub
The home page, which displays all available recommendations in a card-based format for easy scanning.
List View
Clicking a recommendation card reveals a detailed list view, grouping related insights for streamlined review.
Detail Panel
Each recommendation has an accompanying panel where users can dive deeper, customize actions, and view the predicted impact of their decisions.
Impact and Reception
Alpha testing generated overwhelmingly positive feedback, with major clients—such as leading streaming and news platforms—expressing enthusiasm about the hub’s potential to save them significant time and effort.
In just the first round of implementation, Active Assist helped customers remove 9.5 million unused permissions, streamline security, and simplify network configurations.
"Active Assist’s Policy Troubleshooter is going to make my life so much easier. I can't begin to tell you how I've suffered from generic error messages in the past. Policy Troubleshooter is exactly what we need to quickly find, understand, and fix policy misconfigurations. "
Paul Friedman, Sr. Security Engineer, Square
"[Active Assist's] IAM Recommender does all the heavy lifting. It was very easy to auto-apply all the recommendations we got."
Abhi Yadav, Sr Cloud Security Engineer, Uber
By project’s end, we had created an intuitive tool that reshaped how users approached cloud management—making it simpler, safer, and more accessible. Working on Active Assist allowed me to deliver a tangible, user-centered solution that makes a meaningful difference in the daily workflows of our users.
Stakeholders would fill in this form
Outside the box solution
As we came closer to launching, we needed about a dozen stakeholders to craft copy for the recommendations specific to their area of expertise. Many stakeholders came back to us flustered about what copy we needed and where it would appear in the GCP.
I quickly used Google Sheets (Google's spreadsheet editor) to mock up a form where stakeholders could fill in all needed fields. It provided examples and character count guidance. As they filled it in, they automatically had a real-time, medium-fidelity view of what their recommendation would look like, in context, in three key places.
This spreadsheet solved for almost all of the stakeholder confusion about gathering copy to launch.
That information would auto-populate over to these 3 tabs so they could see what their copy looked like in action
Finishing Touches
Once we launched, Google's marketing team put out this video to help users get the most out of the new Recommendations Hub. It does a nice job of summarizing our shipped product.
What's next?
