Humanizing Artificial Intelligence (AI): Better UX in Copilot-generated Power Apps

However, as we’ve delved deeper into the realm of AI, we’ve discovered that this is not always the case. Microsoft’s choice to name their virtual assistant “Copilot” underscores the importance of working alongside this technology to achieve the best results.

While Copilot is an invaluable tool for speeding up the solution design process, its auto-generated designs still benefit greatly from a nuanced human touch to apply the empathy and intuition that great user experience (UX) design demands. Until AI can fully consider and implement this perspective, there are key UX opportunities where your strategic intervention adds significant value.

Solving the right problems

Nielsen Norman Group, the world leaders in user research, are famous for their quote “You are not the user.” Far too often, organizations assume they understand user needs without fully grasping the business processes or constraints, skipping critical discussions before jumping into solution design.

As Jakob Nielsen aptly states, “Even the best designers produce successful products only if their designs solve the right problems. A wonderful interface to the wrong features will fail.”

Copilot cannot tell us what our colleagues need to solve or why, so it is critical we begin each project with some user research. Beginning any app design with user research will not only ensure you are focused on solving the right problems, but it will also help identify 85% of usability problems early in your development process, improve user satisfaction and adoption rates, and reduce the risk of solution failure.

Simplifying navigation and information architecture 

The most satisfying applications do not need a wizard or training guide. They are intuitive to use from the first interaction without any guidance. This intuitive usability is often achieved through well-designed navigation and information architecture, which provide the initial waypoints for what users can do in an app.

While Copilot adheres to logical schemas when generating app components, it sometimes fails to capture users’ mental models. A mental model is an internal representation of how a user believes a system works, which guides their interactions and expectations when using the system. Since Copilot does not conduct user research at the beginning of the process, it misses this critical piece of understanding the user.

To bridge this gap and ensure that every pathway is intuitive, creating a journey where the user’s destination is always clear, it is essential to conduct thorough user research. This research may include interviews, contextual inquiry, card sorting, and usability testing to develop an understanding of your unique and diverse user base. By aligning the design with users’ mental models and expectations, you can create applications that are not only functional but also provide a seamless and satisfying user experience.

Transforming clutter into deliberate simplicity 

Copilot can quickly populate a screen with various controls, buttons, and data visualizations. However, in its quest to deliver functionality, it may overlook the subtle art of careful curation, which focuses on providing the right content in the right context at the right time. You play a crucial role in refining these outputs, stripping away the nonessential items to ensure the visual hierarchy supports your areas of focus and calls to action.

Providing immediate and purposeful feedback 

The 10 Usability Heuristics are a design framework for creating user-friendly interfaces that enhance the overall user experience by following established principles of usability. While all are critical to apply in your Power App design, #1 Visibility of System Status, can have one of the greatest impacts on your user satisfaction.

Have you ever been waiting for a train or bus that was late with no communication? Each minute it is late seems to drag on longer and longer as your stress spikes. If you knew when it was showing up, you could communicate and plan accordingly.

What about when you submit paperwork, receive no feedback, and then find out there are issues with your paperwork when you initiate follow-up after not hearing anything for weeks? Why didn’t anyone tell you there was an issue? Users universally need constant feedback. Copilot-generated interfaces may include basic spinners or progress meters, but these elements fall short of genuine feedback.

Effective design uses immediate, purposefully crafted feedback to inform users of current actions and next steps, transforming uncertainty into assurance. Inline form editing will help reduce form submission errors by 22%, and apps that provide clear system updates will see a 20% increase in user satisfaction. Invest the time to create more descriptive labels and detailed progress, and include time estimates.

Establishing consistent visual and interaction patterns 

Users expect similar elements to behave in similar ways, whether these elements exist only in your solutions or they have seen them in the real world outside your work context. It is a better practice for an organization to establish and adhere to a consistent design system, ensuring that solutions across Microsoft 365 reflect a unified design language.

This consistency not only elevates the aesthetic but also builds trust, as users feel they’re interacting with a thoughtfully orchestrated product. Copilot does not have the ability to consider the Brand Center and any established guidance you have developed, so these elements will need to be applied manually.

Integrating accessibility for all users 

Copilot doesn’t automatically ensure that designs meet the nuanced requirements of all users, particularly those with disabilities. After your designs are initially generated, you can include high-contrast themes, meaningful alternative text, and keyboard navigation, ensuring inclusivity beyond what generative prompts initially provide.

Balancing visibility with workflow zen in modal dialogs 

Popups and modal dialogs can be powerful design tools, though dangerous when overused. Copilot might deploy them to communicate critical information swiftly, but unchecked, they can interrupt user flow and create frustration. Use your best judgment to balance visibility with subtlety, using inline notifications for routine updates and reserving modals for important disruptions.

Crafting clear and constructive error messaging

When things go wrong, communication is critical. Copilot’s error messages sometimes read like diagnostic logs; technical and terse. Transform error messaging into a conversation that is both instructive and clear, guiding users on how to fix issues in plain, actionable language.

Enabling personalization

The varying roles at your organization lead to diversity in needs and on top of that, people have a wide range of personal preferences in how they work. Copilot-generated designs lack the flexibility to accommodate varied workflows and personal preferences. Refine these outputs to add personalization when needed, ensuring that each user feels uniquely catered to with features like role-based customization showing only the features and data relevant to that user.

Embracing iterative refinement and real-user testing 

Now that we have reviewed our Copilot-generated app for UX improvements, we want to test our solution with our end users. While Copilot can generate a functional starting point, the magic of user experience unfolds through rigorous, iterative testing with real people. Use insights from actual usage to reveal subtle friction points and hidden opportunities, transforming a rough draft into a refined masterpiece. According to the Nielsen Norman Group, usability testing with just five users can uncover approximately 85% of usability problems!

While Copilot accelerates the creation of functional prototypes, true solution excellence demands thoughtful, empathetic refinement through understanding and iterating upon user experiences. Until AI can conduct comprehensive user research and deeply understand and respond to user needs with genuine human-centered insights, it will not achieve the depth of understanding needed to create truly user-centered experiences. You play a vital role in bridging the gap between technology and the end user, ensuring that every Power App we build is not only operationally efficient but also genuinely delightful to use.