As AI capabilities like large language models and generative AI continue rapidly advancing, there is a growing need for new design approaches to unlock these powerful technologies' full potential. At a recent Gartner GenAI Sig, Momentum Design Lab, HTEC, and Intuit shared cutting-edge perspectives on "cognitive design" - an emerging methodology for creating software experiences that better understand, anticipate, and respond to user needs and mental models.

Each expert provided valuable perspectives on GenAI's future. Let's examine their insights.

Introducing Cognitive Design  

David Thomson, Founder & CEO of Momentum Design Lab, introduced the cognitive design framework as an evolution beyond traditional "design thinking" processes - one purposely crafted to address the non-linear, multi-dimensional, and open-ended nature of AI interactions. David explained, "We came up with the cognitive design framework because we were trying to find a different way of approaching product design. It's actually not very different from design thinking – think of it as design thinking plus. We use a lot of the same baseline methodologies, but we do many different things around the way people process thoughts rather than the linear task processes you would use with jobs to be done."  

As David explained, their existing design processes broke down when working with generative AI's fluid, adaptive capabilities, stating, "We actually struggled with a couple of early products involving generative AI because we couldn't figure out how to design something with fluidity in it. We are so used to designing things that have a start and end point. With generative AI, you kind of have an open canvas allowing the ability to navigate across multiple dimensions and modalities without restriction, which means you have to start addressing things in the way that people think."  

Designing for Ambient AI  

So, what does this cognitive design approach look like in practice? Peter McNulty, Head of Experience Design at Momentum, described their work on an autonomous vehicle experience that extended well beyond just optimizing navigation from point A to B or providing entertainment content recommendations. Peter explained, "We started by asking, what if all types of data were available? We were looking at ways to extend the experience holistically. Sure, there's mapping, guidance on navigating the environment, and access to information, but we wanted to explore beyond that. What if the vehicle experience could be just part of the user's connected fabric? We believe this is now possible, and we just need to unlock and figure out how to connect it to that time and need.  

Peter calls this approach "ambient design," which accounts for intelligent, contextual interactions that flow seamlessly across modes and dimensions as the user's needs and context change. "I've been starting to use the term' ambient design.' It is multi-touch experiential thinking, so as a design leader, I have been challenging my teams to think this way-you must account for those unknowns. When I said 'beyond,' that is what I was referring to - the context and data are there, but how do we anticipate that need because it is so abstract?

Cognitive design is at the forefront of this shift, transitioning from task-oriented systems to sophisticated, human-like advisors that understand and respond to complex needs. This evolution, coupled with the emergence of ambient computing- characterized by devices adapting to and learning from our behaviors - lays the groundwork for a future where technology is an intuitive extension of ourselves. Ambient design promises a world where technology proactively fulfills our needs, making everyday life more efficient and enjoyable.

Embracing Cognitive Diversity  

For enterprise software applications, designing around AI's cognitive capabilities means deeply understanding and addressing the diverse implicit cognitive processes, mental models, and problem-solving approaches users bring to their tasks.  

As Sava Marinkovich, Head of Growth for Cognitive AI at HTEC Group, explained, successfully applying generative AI requires modeling and designing around both the artificial and human cognitive domains:  

"When you look at the timeline of design frameworks, you have to think of generative AI in two aspects: one is the cognition that comes out of these agents and models, and then the data within them. And when you're structuring how to find these leverage points to apply these generative AI models into design aspects, you have to mirror that against the user's own cognition - the tasks that you're trying to enable them to accomplish."  

Sava highlighted how cognitive design can help identify and fill in the unique cognitive gaps and needs each user brings to a particular task or workflow. He explained, "What are those gaps in cognition - is the task at hand requiring all seven buckets of short-term memory, for example? Or is there something in the logical approach they're taking toward solving a certain task that you identify so you can adapt the model to that individual person? When you expand across individual cognition at the enterprise level or for millions of users, and you tailor the approach for each user, you can start identifying patterns of what cognitions are being used on a massive scale versus an individual scale."  

Unpacking Mental Models  

Joe Preston, VP of Product & Design at Intuit, emphasized the importance of conducting in-depth research to truly unpack users' diverse mental models and cognitive approaches for different tasks. Joe adds, "I think it becomes even more important now to train staff - design staff, product management, research staff - on actually spending time documenting the mental models of people when they're conducting those tasks and really understanding and unpacking that. That is a crucial missing piece that is not done at scale."  

Joe gave the example of how Intuit applied its "Intuit Assist" generative AI capabilities to provide personalized advice and guidance around complex personal finance decisions. He explained, "In the credit card recommendation space, it learns the voice and tone in which the person wants to be spoken to. It can access their essential financial data if they have connected their bank account. And so, when it makes a recommendation, it can explain it in a way that feels extremely personal - it feels as if you are literally on the phone with someone who has had a conversation with you."  

Embedding Cognitive Design  

The experts agreed that successfully embedding cognitive AI requires fundamentally rethinking how design, engineering, data science, and other disciplines collaborate. As David Thomson described for one consulting engagement, "Designers are working hand-in-hand with solutions architects, data scientists, and data engineers to figure out what the model should be. We work quite closely with our engineering teams to make it happen."  

At the same time, Peter McNulty emphasized the inherent challenges of operationalizing such multi-disciplinary cognitive design processes; he adds, "Yes, we challenge the engineering teams - it's been a healthy back and forth, as we are one team, and it's part of the solution process. Overall, they responded that they were very excited about what was possible. Still, we have hit limitations based on the availability of data and all the different technologies involved. We have been taking many iterative approaches, which has been challenging for the project teams because they have already solved the ideal solution. Still, the first iteration of that experience is not fully realized yet."  

Ethical Boundaries  

As the power of cognitive AI systems grows, ensuring ethical, responsible development that avoids manipulative or exploitative experiences is an ever-present concern. Gartner's Andrew Frank raised an issue, "When you think about cognitive design, and especially in use cases like a sales co-pilot or something similar, how do you consider drawing the line between helping people or applications by influencing them versus becoming manipulative and crossing the line that the AI Act is trying to establish around exploiting people's cognitive weaknesses or biases? That must be an issue that arises, right?"  

David Thomson responded that this ethical imperative is front and center for cognitive design, adding, "Absolutely, that's our biggest concern...As designers, we have to make the choice of doing the right thing. Because we are on the forefront, making decisions on how the product is supposed to work...Responsible AI and cognitive design are intertwined; they go through the entire development process from inception all the way to post-release monitoring and frequent user testing."  

David also emphasized the critical importance of rigorously monitoring and receiving user feedback to ensure AI systems don't start to "drift" in misaligned or manipulative directions, especially when using opaque "black box" models. David explains further, "We don't know where an AI system will go because once it's released if it's a black box system, you have to have a feedback loop in place where you're testing it and ensuring that users are utilizing it in the right way - that they are not being manipulated. We don't know what the black box is doing - is it taking on the best qualities from its training or the worst qualities of what it was trained on? We have to be able to ensure the users' safety."  

Future Frontiers  

Looking to the future, the experts expressed significant excitement about cognitive AI enabling seamless, multi-modal "ambient" experiences that fluidly span conversational, visual, and other modalities based on contextual user needs. David Thomson believes we're just at the beginning. He states, "I can guarantee you that both Google, Meta, and Apple have projects underway that we don't even know about yet. I think we're going to see a lot more of that fluidity emerging, where one seamlessly shifts between devices, modalities, and environments."  

On the other hand, Peter McNulty is enthusiastic about designing for such "superpowers" that ambient AI capabilities promise to extend human cognition. He adds, "I'm super excited about really elevating the thoughts and how designers work - storytelling, problem-solving, being the guide to facilitate navigating these big sticky problems that now we have the technology to create these types of experiences."  

Joe Preston envisions significant enterprise productivity gains from "compositional" AI models that can automate mundane tasks at scale through personalized generative pipelines. He adds that one of the most practical applications we'll see is composable models at scale, where you will dramatically reduce the cost to maintain and bifurcate on an uber-personalized scale tied to the design system. He explains, "We spent an exorbitant amount on resources effectively reproducing something that is fairly commoditized. I think there will be a re-education process that must occur throughout the industry, particularly in enterprise software.  

Sava Marinkovich summarized the profound impact of cognitive AI on how organizations approach problem-solving. He said, "What gets me most excited is the complete rethink of how we solve problems. I think it will change how things are organized—from how to structure teams to processes to operating in adapting and fluid environments. We can dramatically change how we deal with tools, what the role of thinking is, and how we approach problems. We're just on the tip of the iceberg of what's coming."  

The cognitive design journey is just beginning. The pioneers and experts at Momentum Design Lab, HTEC, and Intuit are showcasing powerful new methods to truly harmonize human and artificial intelligence. By purposefully modeling and designing for the interplay of human and machine cognition, incredible new "ambient" AI experiences are taking shape.  

Designers get ready—the future of intelligent experiences awaits!

You can learn more about Momentum's Cognitive Design framework in the whitepaper.

Want to speak with an expert instead? Contact us to schedule a time.

Back to top