September 21, 2026
—
5 minutes

Rethinking How We Design and Innovate in an Agentic AI World: Key Takeaways from DDX

‍Was human-centered design all wrong?

‍

No, but it fails to capture the whole picture.

‍

Don Norman introduced human-centered design (HCD), the approach of putting people and their needs at the center of the design process, in the 1980s and almost 50 years later at DDX San Diego 2026, he challenged the philosophy.

‍

Human-centered design gave the industry an important shift in perspective, paying close attention to the people who would actually live with what we create, but there’s tension. Design can create tremendous value for people while also creating unintended consequences for cultures, communities, organizations, and the world around them.

‍

If we optimize for the person in front of us, what happens to everyone and everything outside the frame?

‍

And as technology becomes increasingly agentic, that broader view becomes even more relevant. We're no longer designing only things that people interact with. We're increasingly designing systems that can act, make decisions, and influence what happens next.

‍

‍

‍

Summary

‍

    • Human-centered design does not capture the whole picture
    • New philosophies emerge including life-centered design, intelligence-centered, human-agent centered design, and agent-centered design
    • Four shifts matter most in the age of agentic AI: design for the system, not just the user; turn frameworks into prototypes; study behavior over stated needs; and translate insights and decisions into the language of the business
    • The future of design is about what you do with what you know
  • ‍

    ‍

    ‍

    Emerging design philosophies

    ‍

    ‍

    Life-Centered Design (LCD) expands the focus from the individual user to the larger living system that the product or service exists within. It seeks to align design decisions with global goals such as the United Nation’s Sustainable Development Goals.

    ‍

    Intelligence-Centered Design (ICD) and Human-Agent Centered Design (H-ACD) are extensions of HCD where AI plays an active role in the design process alongside human intelligence, each learning from, making decisions, and mediating the process.

    ‍

    Agent-Centered Design (ACD) puts AI agents at the center as the primary participant in the process, agents building for agents.

    ‍

    ‍

    ‍

    Key elements of design and innovation in an agentic world

    ‍

    ‍

    Start with the people, not the product

    ‍

    ‍

    Good experiences have always started with deep understanding of the people who use the product, service, or system. Needs, expectations, and priorities are shaped by environment, so what works in one market, organization, or culture can't automatically be treated as a universal truth for all.

    ‍

    That's why ethnography and other forms of contextual research remain so valuable. Before asking what we should build, we need to understand the world people are navigating.

    ‍

    Questions to ask:

    ‍

    • Who are we really designing for?
    • What assumptions are we bringing from our own environment?
    • What would happen if this design enters a different context?
    • What are users' stated choices and how does that differ from their actual behavior?

    ‍

    The discipline of understanding people hasn't gone anywhere. It's just being asked to stretch further, to the systems, risks, and impacts beyond individuals.

    ‍

    ‍

    ‍

    Design for the system around the human

    ‍

    ‍

    Behavior is shaped by systems, organizational structures, constraints, and the decisions made around you, so those factors are part of what we are designing for. Agentic AI makes this deep understanding of context even more important.

    ‍

    When AI can act, make decisions, and influence what happens next, it becomes a key part of the system. Therefore, design has to account for the relationship between humans and their agents including how they work together, what artifacts are created along the way, how do they communicate, what decisions does AI make and what are its responsibilities.

    ‍

    Questions to ask:

    ‍

    • What happens outside the immediate experience?
    • How might the system change over time?
    • How do decisions get made? What's the logic behind them?

    ‍

    One practical takeaway is to spend more time understanding what people actually do, not just what they say they do. Diary studies, observation, prototype testing, and live site testing can all help surface what people may not be able to articulate themselves. Rather than simply asking “why,” ask, “What drove you to make that decision?” This is an open question to prompt an honest explanation of what influenced their thinking and led them to take a certain action.

    ‍

    ‍

    ‍

    Do more "design doing"

    ‍

    ‍

    Design has no shortage of frameworks, methodologies, books, and white papers. While these give us useful ways to think, real change only happens when teams transform what they’ve learned into something tangible. In the end, impact isn't measured by the depth of your knowledge, but by what you actually create with it.

    ‍

    Question to ask: What's the smallest thing I could do to build or test this idea this week?

    ‍

    A simple prototype will reveal insights that hours of discussion never could. A real interaction exposes the hidden gaps between what people say and what they actually do. A quick experiment tells a team far more than another round of debate.

    ‍

    At its best, good design thinking must always lead to good "design doing."

    ‍

    ‍

    ‍

    Earn influence by translating value into the language of the business

    ‍

    ‍

    Research and design often sits in the middle of an organization. Leadership sets direction, research gathers insight, design shapes the experience, then implementation follows. When each function speaks a different language, intent can get lost along the way.

    ‍

    Designers and researchers are inherently connectors. We bring together different perspectives, make sense of complexity, and help people see how decisions affect one another. But we aren't often in the room when the decisions themselves are being made. Policy, process, business models, and organizational structures are all forms of design, yet designers are often brought in after those choices have already been made.

    ‍

    There is an opportunity and a need for design thinking and "design doing" to move further upstream, and that starts with speaking the language of the people making those decisions: finances and margins.

    ‍

    Questions to ask:

    ‍

    • What does this insight or design mean for productivity, adoption, risk, growth, for the people working inside the system and for the people affected by it?
    • What would this save, generate, or protect, in dollars, time, or risk?
    • Can we connect the human story to a number?
    • How can I socialize this upwards in my organization?

    ‍

    Business decisions are design decisions.

    ‍

    ‍

    ‍

    Designing for what comes next

    ‍

    ‍

    The next era of design is expanding what we consider when we design.

    ‍

    As systems become more autonomous, designers have an opportunity to shape not only experiences, but the policies, services, and the systems around them.

    ‍

    The future of design is less about finding the right framework and more about what we choose to do with what we know.

    ‍

    Our monthly newsletter, User Tested is for product, design, and research teams who want to make UX research a strategic accelerator.  Subscribe to User Tested.

    ‍

    Share this article
    Author
    Optimal
    Workshop

    Related articles

    View all blog articles
    Header graphic for the article 'Top Tasks in UX: How to Identify What Really Matters...'
    Learn more
    1 min read

    Top Tasks in UX: How to Identify What Really Matters to Your Users

    All the way back in 2014, the web passed a pretty significant milestone: 1 billion websites. Of course, fewer than 200 million of these are actually active as of 2019, but there’s an important underlying point. People love to create. If the current digital age that we live in has taught us anything, it’s that it’s never been as easy to get information and ideas out into the world.

    ‍

    Understandably, this ability has been used – and often misused. Overloaded, convoluted websites are par for the course, with a common tactic for website renewal being to simply update them with a new coat of paint while ignoring the swirling pile of outdated and poorly organized content below.

    ‍

    So what are you supposed to do when trying to address this problem on your own website or digital project? Well, there’s a fairly robust technique called top tasks management. Here, we’ll go over exactly what it is and how you can use it.

    ‍

    Getting to grips with top tasks

    ‍

    Ideally, all websites would be given regular, comprehensive reviews. Old content could be revisited and analyzed to see whether it’s still actually serving a purpose. If not, it could be reworked or just removed entirely. Based on research, content creators could add new content to address user needs. Of course, this is just the ideal. The reality is that there’s never really enough time or resource to manage the growing mass of digital content in this way. The solution is to hone in on what your users actually use your website for and tailor the experience accordingly by looking at top tasks.

    ‍

    What are top tasks? They're basically a small set of tasks (typically around 5, but up to 10 is OK too) that are most important to your users. The thinking goes that if you get these core tasks right, your website will be serving the majority of your users and you’ll be more likely to retain them. Ignore top tasks (and any sort of task analysis), and you’ll likely find users leaving your website to find something else that better fits their needs.

    ‍

    The counter to top tasks is tiny tasks. These are everything on a website that’s not all that important for the people actually using it. Commonly, tiny tasks are driven more by the organization’s needs than those of the users. Typically, the more important a task is to a user, the less information there is to support it. On the other hand, the less important a task is to a user, the more information there is. Tiny tasks stem very much from ‘organization first’ thinking, wherein user needs are placed lower on the list of considerations.
    ‍

    According to Jerry McGovern (who penned an excellent write-up of top tasks on A List Apart), the top tasks model says “Focus on what really matters (the top tasks) and defocus on what matters less (the tiny tasks).”

    ‍

    How to identify top tasks

    ‍

    Figuring out your top tasks is an important step in clearing away the fog and identifying what actually matters to your users. We’ll call this stage of the process task discovery, and these are the steps:

    ‍

    1. Gather: Work with your organization to gather a list of all customer tasks
    2. Refine: Take this list of tasks to a smaller group of stakeholders and work it down into a shortlist
    3. User feedback: Go out to your users and get a representative sample to vote on them
    4. Finalise: Assemble a table of tasks with the one with the highest number of votes at the top and the lowest number of votes at the bottom

    ‍

    We’ll go into detail on the above steps, explaining the best way of handling each one. Keep in mind that this process isn’t something you’ll be able to complete in a week – it’s more likely a 6 to 8-week project, depending on the size of your website, how large your user base is and the receptiveness of your organization to help out.

    ‍

    Step 1: Gather – Figure out the long list of tasks

    ‍

    The first part of the task process is to get out into the wider organization and discover what your users are actually trying to accomplish on your website or by using your products. It’s all about getting into the minds of your users – trying to see the world through their eyes, effectively.

    ‍

    If you’re struggling to think of places where you might find customer tasks, here are some of the best sources:

    ‍

    • Analytics: Take a deep dive into the analytics of your website or product to find out how people are using them. For websites, you’ll want to look at pages with high traffic and common downloads or interactions. The same applies to products – although the data you have access to will depend on the analytics systems in place.
    • Customer support teams: Your own internal support teams can be a great source of user tasks. Support teams commonly spend all day speaking to users, and as a result, are able to build up a cohesive understanding of the types of tasks users commonly attempt.
    • Sales teams: Similarly, sales teams are another good source of task data. Sales teams typically deal with people before they become your users, but a part of their job is to understand the problems they’re trying to solve and how your website or product can help.
    • Direct customer feedback: Check for surveys your organization has run in the past to see whether any task data already exists.
    • Social media: Head to Twitter, Facebook and LinkedIn to see what people are talking about with regards to your industry. What tasks are being mentioned?

    ‍

    It’s important to note that you need to cast a wide net when gathering task data. You can’t just rely on analytics data. Why? Well, downloads and page visits only reflect what you have, but not what your users might actually be searching for.

    ‍

    As for search, Jerry McGovern explains why it doesn’t actually tell the entire story: “When we worked on the BBC intranet, we found they had a feature called “Top Searches” on their homepage. The problem was that once they published the top searches list, these terms no longer needed to be searched for, so in time a new list of top searches emerged! Similarly, top tasks tend to get bookmarked, so they don’t show up as much in search. And the better the navigation, the more likely the site search is to reflect tiny tasks.”

    ‍

    At the end of the initial task-gathering stage you should be left with around 300 to 500 tasks. Of course, this can scale up or down depending on the size of the website or product.

    ‍

    Step 2: Refine – Create your shortlist

    ‍

    Now that you’ve got your long list of tasks, it’s time to trim them back until you’ve got a shortlist of 100 or less. Keep in mind that working through your long list of tasks is going to take some time, so plan for this process to take at least 4 weeks (but likely more).

    ‍

    It’s important to involve stakeholders from across the organization during the shortlist process. Bring in people from support, sales, product, marketing and leadership areas of the organization. In addition to helping you to create a more concise and usable list, the shortlist process helps your stakeholders to think about areas of overlap and where they may need to work together.

    ‍

    When working your list down to something more usable, try and consolidate and simplify. Stay away from product names as well as internal organization and industry jargon. With your tasks, you essentially want to focus on the underlying thing that a user is trying to do. If you were focusing on tasks for a bank, opt for “Transactions” instead of “Digital mobile payments”. Similarly, bring together tasks where possible. “Customer support”, “Help and support” and “Support center” can all be merged.

    ‍

    At a very technical level, it also helps to avoid lengthy tasks. Stick to around 7 to 8 words and try and avoid verbs, using them only when there’s really no other option. You’ll find that your task list becomes quick to navigate when tasks begin with “look”, “find” and “get”. Finally, stay away from specific audiences and demographics. You want to keep your tasks universal.

    ‍

    Step 3: User feedback – Get users to vote

    ‍

    With your shortlist created, it’s time to take it to your users. Using a survey tool like Optimal's Surveys, add in each one of your shortlisted tasks and have users rank 5 tasks on a scale from 1 to 5, with 5 being the most important and 1 being the least important.

    ‍

    If you’re thinking that your users will never take the time to work through such a long list, consider that the very length of the list means they’ll seek out the tasks that matter to them and ignore the ones that don’t.

    ‍

    A section of the customer survey in Questions.
    A section of the customer survey in Questions.

    ‍

    Step 4: Finalize – Analyze your results

    ‍

    Now for the task analysis side of the project. What you want at the end of the user survey end of the project is a league table of entire shortlist of tasks. We’re going to use the example from Cisco’s top tasks project, which has been documented over at A List Apart by Gerry McGovern (who actually ran the project). The entire article is worth a read as it covers the process of running a top task project for a large organization.

    ‍

    Here’s what a league table of the top 20 tasks looks like from Cisco:

    ‍

    A league table of the top 20 tasks from Cisco’s top tasks project.
    A league table of the top 20 tasks from Cisco’s top tasks project. Credit: Jerry McGovern.

    ‍

    Here’s the breakdown of the vote for Cisco’s tasks:

    • 3 tasks got the first 25 percent of the vote
    • 6 tasks got 25-50 percent of the vote
    • 14 tasks got 50-75 percent of the vote
    • 44 tasks got 75-100 percent of the vote

    ‍

    While the pattern may seem surprising, it’s actually not unusual. As Jerry explains: “We have done this process over 400 times and the same patterns emerge every single time.”

    ‍

    Final thoughts

    ‍

    Focusing on top tasks management is really a practice that needs to be conducted on a semi-regular basis. The approach benefits organizations in a multitude of ways, bringing different teams and people together to figure out how to best address why your users are coming to your website and what they actually need from you.

    ‍

    As we explained at the beginning of this article, top tasks is really about clearing away the fog and understanding on what really matters. Instead of spreading yourself thin and focusing on a host of tiny tasks, hone in on those top tasks that actually matter to your users.

    ‍

    Understanding how to improve your website

    ‍

    The top tasks approach is an effective way of giving you a clear idea of what you should be focusing on when designing or redesigning your website, but this should really just be one aspect of the work you do.

    ‍

    Utilizing a host of other UX research methods can give you a much more comprehensive idea of what’s working and what’s not. With card sorting, for example, you can learn how your users think the content on your website should be arranged. Then, with this data in hand, you can use tree testing to assemble draft structures of your website and test how people navigate their way through it. You can keep iterating on these structures to ensure you’ve created the most user-friendly navigation.
    ‍

    Take a look at our 101 guides to learn more about card sorting and tree testing, as well as the other user research methods you can use to make solid improvements to your website. If you’d rather just start putting methods into practice using user research tools, take our UX platform for a spin for free here.

    Header graphic for the article 'Addressing AI Bias in UX: How to Build Fairer Digital...'
    Learn more
    1 min read

    Addressing AI Bias in UX: How to Build Fairer Digital Experiences

    The Growing Challenge of AI Bias in Digital Products

    ‍

    AI is rapidly reshaping our digital landscape, powering everything from recommendation engines to automated customer service and content creation tools. But as these technologies become more widespread, we're facing a significant challenge: AI bias. When AI systems are trained on biased data, they end up reinforcing stereotypes, excluding marginalized groups, and creating inequitable digital experiences that harm both users and businesses.

    ‍

    This isn't just theoretical, we're seeing real-world consequences. Biased AI has led to resume screening tools that favor male candidates, facial recognition systems that perform poorly on darker skin tones, and language models that perpetuate harmful stereotypes. As AI becomes more deeply integrated into our digital experiences, addressing these biases isn't just an ethical imperative t's essential for creating products that truly work for everyone.

    ‍

    ‍

    ‍

    Why Does AI Bias Matter for UX?

    ‍

    For those of us in UX and product teams, AI bias isn't just an ethical issue it directly impacts usability, adoption, and trust. Research has shown that biased AI can result in discriminatory hiring algorithms, skewed facial recognition software, and search engines that reinforce societal prejudices (Buolamwini & Gebru, 2018).

    ‍

    When AI is applied to UX, these biases show up in several ways:

    ‍

    • Navigation structures that favor certain user behaviors
    • Chatbots that struggle to recognize diverse dialects or cultural expressions
    • Recommendation engines that create "filter bubbles" 
    • Personalization algorithms that make incorrect assumptions 

    ‍

    These biases create real barriers that exclude users, diminish trust, and ultimately limit how effective our products can be. A 2022 study by the Pew Research Center found that 63% of Americans are concerned about algorithmic decision-making, with those concerns highest among groups that have historically faced discrimination.

    ‍

    ‍

    ‍

    The Root Causes of AI Bias

    ‍

    To tackle AI bias effectively, we need to understand where it comes from:

    ‍

    1. Biased Training Data

    ‍

    AI models learn from the data we feed them. If that data reflects historical inequities or lacks diversity, the AI will inevitably perpetuate these patterns. Think about a language model trained primarily on text written by and about men,  it's going to struggle to represent women's experiences accurately.

    ‍

    2. Lack of Diversity in Development Teams

    ‍

    When our AI and product teams lack diversity, blind spots naturally emerge. Teams that are homogeneous in background, experience, and perspective are simply less likely to spot potential biases or consider the needs of users unlike themselves.

    ‍

    3. Insufficient Testing Across Diverse User Groups

    ‍

    Without thorough testing across diverse populations, biases often go undetected until after launch when the damage to trust and user experience has already occurred.

    ‍

    ‍

    ‍

    How UX Research Can Mitigate AI Bias

    ‍

    At Optimal, we believe that continuous, human-centered research is key to designing fair and inclusive AI-driven experiences. Good UX research helps ensure AI-driven products remain unbiased and effective by:

    ‍

    Ensuring Diverse Representation

    ‍

    Conducting usability tests with participants from varied backgrounds helps prevent exclusionary patterns. This means:

    ‍

    • Recruiting research participants who truly reflect the full diversity of your user base
    • Paying special attention to traditionally underrepresented groups
    • Creating safe spaces where participants feel comfortable sharing their authentic experiences
    • Analyzing results with an intersectional lens, looking at how different aspects of identity affect user experiences
      ‍

    ‍

    Establishing Bias Monitoring Systems

    ‍

    Product owners can create ongoing monitoring systems to detect bias:

    ‍

    • Develop dashboards that track key metrics broken down by user demographics
    • Schedule regular bias audits of AI-powered features
    • Set clear thresholds for when disparities require intervention
    • Make it easy for users to report perceived bias through simple feedback mechanisms
      ‍

    ‍

    Advocating for Ethical AI Practices

    ‍

    Product owners are in a unique position to advocate for ethical AI development:

    ‍

    • Push for transparency in how AI makes decisions that affect users
    • Champion features that help users understand AI recommendations
    • Work with data scientists to develop success metrics that consider equity, not just efficiency
    • Promote inclusive design principles throughout the entire product development lifecycle
      ‍

    ‍

    ‍

    The Future of AI and Inclusive UX

    ‍

    As AI becomes more sophisticated and pervasive, the role of customer insight and UX in ensuring fairness will only grow in importance. By combining AI's efficiency with human insight, we can ensure that AI-driven products are not just smart but also fair, accessible, and truly user-friendly for everyone. The question isn't whether we can afford to invest in this work, it's whether we can afford not to.

    ‍

    Learn more
    1 min read

    What Our AI Interviewer Taught Us About Conversational AI

    AI-moderated interviews are moving from experiment to everyday workflow. But for anyone who cares about user experience, the real question isn’t whether an AI can ask good questions. It’s whether the person on the other side wants to answer them. And that comes down to the design more than the model.

    ‍

    People are more comfortable talking to AI than you might expect. In an Adobe Express survey of 1,005 US consumers, 61% said they were comfortable having conversations with AI-powered systems. A COPC study across six countries found people who knew they were talking to AI reported 34% higher satisfaction than those who weren’t told.

    ‍

    But a quick chat with a bot isn’t a research interview. So we put it to the test. We ran a study using Optimal’s AI Interviewer with participants at different levels of AI and UX experience, including some who were skeptical going in (and rightfully so!). Because if an AI interview only works for people who already love AI, it doesn’t really work.

    ‍

    ‍

    What participants told us

    ‍

    Most participants described the experience as seamless, smooth, natural and even fun, including several who had reservations going in. Three things stood out.

    ‍

    1. It felt like a conversation, not a script

    ‍

    Participants noticed that follow-ups were built on what they had just said. One put it simply:

    ‍

    “I really felt like I was talking to a real person.”

    ‍

    Another appreciated that the AI summarised their points before moving on, instead of asking the same question in different ways. That was a frustration they’d had with other AI interview tools.

    ‍

    2. Less pressure made it easier to open up

    ‍

    Several participants found the AI easier to talk to than a live interviewer. Without a person across the table, there was less pressure to perform.

    ‍

    “I have no problems speaking freely.”

    ‍

    On-screen text alongside the spoken conversation also mattered. One participant with auditory processing difficulties called it out as a genuinely useful accessibility feature.

    ‍

    Not everyone was fully won over. One participant felt a human would better grasp the emotion behind their answers, though they still rated the experience positively.

    ‍

    3. Choice built confidence

    ‍

    Participants could pick how the conversation flowed. One chose “submit response” mode, deciding when each answer was finished. After a good first experience, they said they’d be happy to try “free form” mode next, where the AI responds naturally once they stop speaking.

    ‍

    ‍

    Want to hear it from participants directly?
    ‍Watch the highlight reels in our full study write-up →

    ‍

    ‍

    What designers can take from this

    ‍

    The takeaway reaches well beyond research tools: whether an AI conversation works depends less on the model and more on the experience designed around it. If you’re designing any conversational AI, these are worth stealing.

    ‍

    • Prove you’re listening. Responses that reference the user’s own words build trust faster than clever phrasing. A short summary before moving on signals “I heard you.”
      ‍
    • Don’t make people repeat themselves. Rephrasing the same question feels robotic. Move the conversation forward instead.
      ‍
    • Lower the stakes. Removing social pressure can make people more candid. Clear instructions and a predictable flow help them settle in.
      ‍
    • Offer more than one channel. Pairing voice with on-screen text supports people who process information differently, and makes the experience easier for everyone.
      ‍
    • Give users control over pacing. Letting people choose when to submit, or when to go free form, turns uncertainty into confidence.
      ‍
    • Be honest about limits. Some people will still prefer a human for emotionally nuanced topics. Design AI as an option, not the only path.

    ‍

    ‍

    See how Optimal’s AI Interviewer works

    ‍

    We built these lessons into Optimal’s AI Interviewer, an AI-moderated research tool designed to work with you, not instead of you. It helps teams run more interviews without losing the depth that makes qualitative research valuable.

    ‍

    1. Add your context. Research goals, audience details and supporting material guide every follow-up.
      ‍
    2. Build your discussion guide. Set sections, questions, goals and how deeply the AI should probe.
      ‍
    3. Let it run the interview. The AI listens, adapts and explores relevant threads while staying on your goals.
      ‍
    4. Get cited insights. An AI-generated summary and key themes, with citations at study, section and question level, plus highlight reels that jump to the moment behind each insight.

    ‍

    Take this two-minute interactive tour to see it for yourself.

    ‍

    ‍

    The bottom line

    ‍

    AI interviews aren’t about replacing the human side of research. They give teams another way to listen, at a scale that used to need a lot more people and time. And when the experience is designed well, participants are happy to talk.

    ‍

    Optimal’s AI Interviewer is now in Early Access, book a demo to learn more.

    Seeing is believing

    Explore our tools and see how Optimal makes gathering insights simple, powerful, and impactful.