“Now it’s like creating a book where the pages are always changing,” says Yannis Paniaras, a principal designer in the Microsoft Digital Studio. “At the critical junction in the UX, where humans interact with Copilot, the AI transforms into the conductor of the user experience. This shift is enabling our designers to move away from defining fixed flows to embracing a non-deterministic design style orchestrated by the AI.”

Microsoft Digital Studio is our team of designers and researchers in Microsoft Digital. The Microsoft Digital Studio team is committed to using their expertise in design, research, content strategy, accessibility, and product planning to create experiences that empower Microsoft employees to achieve more in their lives.

Paniaras has observed that designing an AI-enabled product is very different from designing a traditional desktop or mobile app. In conversations with designers, program managers, and developers, he frequently encounters questions about how various product-making disciplines should coordinate their work in this new context.

“We have Copilot, powered by a large language model (LLM), and we use Fluent AI design language for experiences that rely on lean graphical user interfaces with dynamic prompts and dynamically generated, contextual cards,” Paniaras says. “These provide just-in-time user interface elements that map to the generative flow. Consequently, designers are shifting their focus from standard UI towards the vocabulary of prompts, dynamically designed adaptive cards, and on finding consistency withing the UX context. These elements are becoming the new building blocks of AI-based UX design.”

The Microsoft Digital Studio team’s designers still work in Figma, the popular design and prototyping tool, but their designs need to remain open-ended and sometimes more abstract, rather than a set of fixed linear designs.

“The design becomes a set of probabilities,” Paniaras says. “While this poses a challenge for designers, it also encourages us to collaborate more closely with everyone else.”

Laura Bergstrom, a principal UX manager for the Unified Employee Experience team, adds that content designers and designers on her team developed guidance for engineers to scale Copilot responses creating consistent, reliable responses with the right tone of voice at the right time.

“With all the power of generative AI, user experience and design are still responsible for the quality of the experience and the outcome, so we’re finding ways to scale working with engineering and data science,” Bergstrom says.

Spurring collaboration with AI

Using AI to quickly align plans and goals is causing a shift in the way the entire product-making crew works together. “All the different disciplines are working together to get things in place,” Paniaras says.

He tells a story of a designer who worked concurrently with PMs and engineers to design prompts, comparing it with the previous way of doing things.

“It used to be different: you had research, based on that you would have ideas, prototype certain things, build them, then engineers would test them,” Paniaras says. “It was more linear.”

Modern engineering with AI requires a shift to a more collaborative culture on product teams, where there aren’t clear lines of ownership and people can work flexibly together. It’s similar to the shift that engineering went through from the waterfall approach to agile, when instead of owning specific pieces, engineers swarmed over one part of the product for a sprint, then swarmed on another part in the next sprint.

Transitional UI interfaces

Victor Albahadly, a senior UX designer on the Microsoft Digital Studio team, says AI has potential to transform the way he does his core job, which is to test to find out where the designs he and his team build breakdown and fail to meet the needs of the people who will use them.

“I need to figure out what the user wants,” Albahadly says. “When we build an application, I need to know where they are coming from, what they want to do, and where the experience that we’re building for them will break down.”

The challenge is that he has to sample the experiences users have with his designs and extrapolate what he learns to the rest of the design. And importantly, he does this at scale—not for just one person, but for all the people who use the application.

“I need to test how the experience will work for many people,” he says. “That’s an intense process.”

AI has the potential to change that because it will be able to see everything—something a human will never be able to do on their own.

“With AI’s help, someday in the near future, I’ll be able to test the entire application,” Albahadly says. “There will be a lot of power in that.”

AI can help designers get this kind of scale at every step in the process, which not only makes the results far more accurate, but also much faster.

Transforming user testing with AI

The future of ideation

Albahadly also envisions AI enhancing parts of the design process. Today, ideation is done by talking to experts and customers, holding brainstorming sessions, and doing workshops. In the future, he suggests he could do similar ideation with his teammates and AI.

“In your app, say there’s a huge drop-off of traffic coming from Japan,” he says. “Now we need to do a workshop to find out why this is happening. The AI could point to specific stuff like a language barrier or culture barrier, or a time issue like a holiday. Instead of taking a week to ideate, it could become a step in the process the same day.”

In addition to changes in design processes, generative AI is changing the user experience.

“We’ve had a linear way of pumping out experiences—an OS, products on top of it, and apps,” Bergstrom says. “Now there are different copilots, different extensibility, ways of doing things on surfaces. This all has to make sense to a user end-to-end.”

It requires a lot of design thinking to produce that experience.

Data quality is also crucial to producing an experience that makes sense. “Generative AI is a wildcard, which requires data to be more pristine,” Bergstrom says.

For example, the LLM for Microsoft 365 can go through all your emails and SharePoint sites. If you type in “benefits,” it should identify the authoritative source and display that information—not go through your email to find every benefits-related message you’ve ever gotten.

Transforming work with AI

What about the potential of AI to do routine, repetitive work and give people the time to do higher value work? Bergstrom sees a wide range of opportunities.

“We can use generative AI to help employees with everyday tasks, from finding the best place to park to managing the immigration process to identifying the best selections for employee benefits,” Bergstrom says. “And for large enterprises, we can use generative AI to help manage facilities by identifying cost-to-benefit ratios, building usage, and for finding the best locations to have offices.”

Both Bergstrom and Albahadly see an opportunity for AI to help employees write their performance reviews. Bergstrom notes that it could help managers combine review feedback from multiple sources and tie it to OKRs.

And Albahadly says that for employees, AI can help with writing their own performance reviews.

“That’s been a challenge for most Microsoft employees, because at the end of the year, you have to sit and remember everything you worked on,” he says.

Because AI will be exposed to your meetings, your calendar, your projects, it will be easy for it to co-write your review with you.

“In the future, it will be less writing and more selecting stuff, and AI will generate a whole year for you,” Albahadly says.

With all this transformation happening, some people worry about the future of work.

Paniaras is optimistic.

“Everything around us, including our roles, work, processes, and definitions of values, has been created by us humans” he says. “Whenever any of these dimensions change, we inevitably end up redefining them or filling the void. But you need to have that thinking attitude, and the recognition that everything around us is a result of our own making.”

Bergstrom agrees.

“Durable problems don’t change,” she says. “But now we have infinitesimally more ways to solve for those problems with an intelligent assistant that can anticipate needs and predicts possibilities—we’re just trying to figure out how to harness all the capability in our designs.”

Try out Microsoft 365 Copilot to learn what you can do with AI.

Watch John Maeda’s LinkedIn Learning class—UX for AI: Design Practices for AI Developers—to learn more about how collaboration works with AI.

Here are some tips for getting started with generative AI at your company:

1. Embrace AI as a collaborator:
   • Consider AI as a creative partner. It can augment your design process by suggesting patterns, layouts, and interactions.
   • Collaborate with AI tools to generate design variations, explore possibilities, and iterate faster.
2. Understand AI’s capabilities and limitations:
   • Familiarize yourself with the types of AI algorithms commonly used in design, such as neural networks, generative adversarial networks (GANs), and reinforcement learning.
   • Recognize that AI has limitations—it can’t replace human intuition, empathy, or domain expertise. Use it as a tool to enhance your creativity.
3. Design for adaptability and personalization:
   • AI-driven UX should be adaptable and personalized. Create interfaces that adjust dynamically based on user behavior, context, and preferences.
   • Use AI to tailor experiences for individual users, providing relevant content and recommendations.
4. Collect and curate data:
   • AI models require data to learn and improve. Collect relevant user data (with privacy considerations) to train AI algorithms.
   • Curate high-quality datasets that represent diverse user scenarios and behaviors.
5. Iterate and refine AI models:
   • Start with simple AI models and gradually increase complexity. Iterate based on user feedback and real-world usage.
   • Regularly evaluate and fine-tune AI models to ensure they align with user needs and business goals.
6. Ethical considerations:
   • Be mindful of biases in AI algorithms. Ensure fairness, transparency, and inclusivity.
   • Understand the ethical implications of AI-driven decisions and design accordingly.
7. Learn from existing AI-driven products:
   • Study successful AI-powered products and services. Analyze how they integrate AI seamlessly into the user experience.
   • Learn from industry leaders and adapt their best practices to your own projects.

Remember, AI is a powerful tool, but it’s most effective when combined with human creativity and empathy. By embracing AI and understanding its role, UX designers can create innovative, personalized, and adaptive experiences for users.

Find out what you can do with Microsoft 365 Copilot.

• Take this UX for AI: Design Practices for AI Developers class with Microsoft’s John Maeda.
• Explore these five communications-based tips for prompting Microsoft 365 Copilot.
• Unpack deploying Microsoft 365 Copilot internally at Microsoft.
• Discover embracing emerging technology at Microsoft with new AI certifications.

• Want more information? Email us and include a link to this story and we’ll get back to you.
• Please share your feedback with us—take our survey and let us know what kind of content is most useful to you.

The post Redesigning how we work at Microsoft with generative AI appeared first on Inside Track Blog.