<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Fatma Dogan Seckin on Medium]]></title>
        <description><![CDATA[Stories by Fatma Dogan Seckin on Medium]]></description>
        <link>https://medium.com/@fatmadoganseckin?source=rss-f427704d22c0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*kBjvms4r6cTHUltGDiQlag.png</url>
            <title>Stories by Fatma Dogan Seckin on Medium</title>
            <link>https://medium.com/@fatmadoganseckin?source=rss-f427704d22c0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 01:08:05 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@fatmadoganseckin/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[5 Claude Code Mods That Make Design-to-Code Work Easier to Review]]></title>
            <link>https://medium.com/space-ux-design/5-claude-code-mods-that-make-design-to-code-work-easier-to-review-22915681716b?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/22915681716b</guid>
            <category><![CDATA[claude-code]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[design-systems]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Thu, 19 Nov 2026 16:31:01 GMT</pubDate>
            <atom:updated>2026-11-19T16:31:01.782Z</atom:updated>
            <content:encoded><![CDATA[<p><em>From following a long build to catching changes in shared components, these add-ons help me see what happened between the Figma design and the working prototype.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-LkklNP1T1Xd1Y2bvQIPiw.png" /><figcaption>From design file to working preview, with the shared component change surfaced for review.</figcaption></figure><p>When I move a design from Figma into a working prototype, the first build is usually the easy part. The slower part starts when I open the preview. The desktop dashboard might look right at first glance, but then I resize the window and notice the filter bar wraps differently from the design. I ask Claude to adjust the spacing, refresh the preview, and spot the same change on a report page I hadn’t planned to touch.</p><p>That’s when I want to understand how the prototype was built. Are both pages using the same filter component? Did Claude adjust the spacing on this page, or did it change a token that several screens use? The answer changes how I give feedback. If the component is shared, I might want to review the change across the product. If the spacing is specific to one layout, I can explain where the adjustment belongs.</p><p>A session summary gives me a list of files that changed. During a longer build, I also want to know where Claude is working, which parts of the prototype are ready to review, and when I need to step in with a design decision. That’s what interested me about Claude Code mods after reading Nick Babich’s piece in UX Planet. <a href="https://uxplanet.org/5-new-claude-mods-that-will-change-how-you-work-618fa55bc7db">UX Planet</a></p><p>Mods are plugins that can change how Claude Code looks and behaves. They can respond to events in a session, such as when you submit a prompt or Claude calls a tool. Some add a panel you can keep open during the build. Others give you a chance to review an action or revisit an edit. Each of these five fits a different moment in my design-to-code workflow. <a href="https://code.claude.com/docs/en/plugins/mods/overview">Claude Code Docs</a></p><h4>1. Savvy Progress helps you pick up where you left off</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LdDd31au6lbOA0f31ticDQ.png" /><figcaption>A progress panel shows which part of the build is ready to review.</figcaption></figure><p>Imagine asking Claude to build an analytics dashboard with a date filter, several report views, and loading and empty states. While it works, I might go back to Figma to refine the empty state copy or check what happens to the chart legend on a narrow screen. By the time I return, there could be several tasks in progress, and I want to know where to start looking.</p><p>Savvy Progress adds a progress bar and a panel that shows what active subagents are working on, along with task progress and context usage. If the overview is ready while the report views are still being built, I can open the preview and review the overview first. I can check whether the date filter is easy to find and whether the chart still makes sense on mobile while the rest of the build continues.</p><p>I’d use the panel to decide what to review next. Once I open the preview, I might find that a report view is in place but its empty state still needs a decision. For example, should someone who hasn’t added a website see an invitation to add one? And should someone whose website is connected see an explanation that data is still coming in? Those are two different moments in the product, even if the screen looks similar at first.</p><h4>2. Filetree shows where a change is happening</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RUMQLGjC28XajgzFqkFgPg.png" /><figcaption>A small dashboard request can reach a shared header used across screens.</figcaption></figure><p>“Make the dashboard header more compact” sounds like a small request. In Figma, I can usually tell whether I’m adjusting an instance on one screen or editing the main component. In code, the same request could lead Claude to edit the dashboard page, update a shared header, or change a spacing token. I’d want to review a different set of screens depending on which route it takes.</p><p>Filetree adds a sidebar that shows which files Claude is reading and editing. I’d keep it open when refining a prototype that already uses a design system. If Claude opens the shared header to understand how it works, I can follow along. If it starts editing that file, I can decide whether the compact treatment belongs on every page that uses the header.</p><p>On an analytics dashboard, I might need a tighter header to give the chart more room on mobile. On a settings page, the existing spacing might still make sense because the title and description need room to wrap. Seeing where the edit is happening gives me a chance to explain that difference before I’m reviewing several screens at once.</p><h4>3. Replay Theater helps you trace a visual regression</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z7gDpIqYWZuusk-00XsSSw.png" /><figcaption>Stepping through edits connects a spacing change to the date picker that moved.</figcaption></figure><p>Say I ask Claude to tighten the spacing in a filter bar. The filter bar looks right after I refresh the preview, but later I notice the date picker has shifted on another screen. If the session also included changes to the page layout and responsive styles, it can take a while to figure out which edit connects the two.</p><p>Replay Theater is one of Anthropic’s sample mods. It lets you step through the edits captured during a session and see what changed in each file. I’d use it to follow the sequence of changes and narrow down when the date picker moved. If the edit affected the shared filter component, I’d know which other screens to open and check. If it changed a spacing token, I’d look at where else the token is used before asking for another adjustment.</p><p>That also helps me give more specific feedback. I can ask Claude to keep the date picker aligned across the reports and adjust the spacing around it on the dashboard. Since the replay captures changes made through Claude’s editing tools, I’d also check the Git diff before wrapping up the review.</p><h4>4. Reflect turns repeated feedback into project guidance</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MRR2IBk9oX7Qi8DfHy-3WA.png" /><figcaption>A review comment becomes a suggested project rule you can save or skip.</figcaption></figure><p>Some feedback belongs to one screen. I might ask Claude to move a chart legend below the graph because the labels wrap at that screen size. Other feedback comes up across the whole product. “Check the existing button variants before creating a new one” is something I’d want Claude to remember when it starts the next feature.</p><p>Reflect can flag corrections that might be worth saving. You can edit or skip a suggestion, or approve it as guidance in CLAUDE.md. I’d use it when I catch myself repeating a decision the team has already made. If we’ve agreed to reuse the existing button variants, I can save that guidance once and refer back to it when we review a future screen.</p><p>The same applies to product behavior. In an analytics workspace, we might decide that a new account should open in an empty workspace with an option to add a website. If I keep correcting prototypes that open with sample charts, I can turn that feedback into a clear project instruction. I’d include what the user should see on the first visit so Claude has enough context to build the right state next time.</p><h4>5. Blast Radius shows what certain commands could affect</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*phJC5DZ38jMYqwHpchpjRg.png" /><figcaption>Before a cleanup command runs, you can see exactly which files are at risk.</figcaption></figure><p>By the end of a prototype sprint, I might have two versions of a dashboard, several sets of exported assets, and components from an experiment I’ve moved on from. At that point, it’s tempting to ask Claude to “clean up the old prototype files” and get back to the design work. The trouble is that “old” can mean different things. An earlier layout might be ready to remove, while an asset exported for that layout could still be used in the current demo.</p><p>Blast Radius is another Anthropic sample mod. It intercepts certain commands and previews what they could affect before you decide whether to proceed. For supported deletion commands, that can mean seeing the files at risk. I’d use the preview alongside the project’s existing permission settings.</p><p>If the list includes a folder I expected to keep, I can pause and ask Claude to check where those files are still referenced. I can then narrow the request to the experiment I meant to remove. That’s easier to do when I can see the paths Claude is about to act on.</p><h4>What I’d try first</h4><p>I’d start with Filetree on my next design-to-code project. A common review comment for me is something like “Give the primary button a little more room on this pricing card.” If Claude starts editing the shared button, I can check whether we already have a variant that fits. I might also realize that the button itself is fine and the extra space belongs in the card layout.</p><p>I make that decision in Figma all the time. Before editing a main component, I think about the other screens that use it and whether they should pick up the change. I want to make that decision just as deliberately when Claude is working in code. If I ask for a one-card adjustment, I’d like to see where it lands while there’s still time to talk through it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=22915681716b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/5-claude-code-mods-that-make-design-to-code-work-easier-to-review-22915681716b">5 Claude Code Mods That Make Design-to-Code Work Easier to Review</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Claude Design Isn’t Dying. AI Design Is Evolving.]]></title>
            <link>https://fatmadoganseckin.medium.com/claude-design-isnt-dying-ai-design-is-evolving-a13d221b3b9f?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/a13d221b3b9f</guid>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-strategy]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Sat, 07 Nov 2026 06:01:01 GMT</pubDate>
            <atom:updated>2026-11-07T06:01:01.884Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f9obfsqeegnXwSbpN0DdsA.png" /></figure><p><em>Why the future of product design is moving beyond interface generation toward integrated workflows, human judgment, and product strategy.</em></p><p>The announcement that Anthropic is shutting down the standalone Claude Design experience has sparked another debate about the future of designers. Some people interpret the decision as evidence that AI-powered design tools have failed, while others believe it proves that creativity and design expertise cannot be automated. Both interpretations make for compelling headlines, but neither fully captures what is happening.</p><p>Claude Design isn’t simply disappearing. Anthropic is moving its design capabilities into the broader Claude Artifacts experience, with the standalone service scheduled to close on December 14, 2026. This is an important distinction because closing a standalone product does not necessarily mean abandoning the technology behind it. Companies regularly consolidate products when they see greater value in integrating capabilities into existing platforms, and AI design tools are increasingly becoming part of broader creative and development environments.</p><p>Rather than treating this announcement as the end of AI design, I see it as an opportunity to reconsider how artificial intelligence fits into modern product development. The industry is gradually moving beyond tools that generate isolated interfaces toward environments that connect research, design, development, and iteration. This transition has implications not only for the tools designers use but also for the skills, responsibilities, and strategic decisions that define their work.</p><h4>Generating Interfaces Is Not the Same as Designing Products</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FKMEyErIcXTW3cT-cttLXg.png" /><figcaption>Interface vs. Product Design</figcaption></figure><p>Artificial intelligence has significantly changed how quickly digital interfaces can be produced. Designers and developers can now generate dashboards, landing pages, interactive prototypes, and reusable components within minutes. Tasks that previously required hours of manual execution can increasingly be accelerated through AI-assisted workflows, making experimentation more accessible and reducing the time between an initial idea and a working interface.</p><p>However, generating an interface is not equivalent to designing a successful product. Product design begins long before the first screen is created and continues well after a product is launched. It involves understanding user needs, identifying meaningful problems, evaluating opportunities, and making decisions that balance usability, business objectives, and technical constraints. A visually impressive interface may demonstrate strong execution, but that does not necessarily mean it addresses a genuine user problem or creates measurable value.</p><p>Consider a financial dashboard generated through AI. The interface might include attractive charts, portfolio summaries, transaction histories, and sophisticated data visualizations. At first glance, the result may appear ready for production, especially if it follows familiar interface patterns and established visual conventions. Yet the quality of the product depends on much deeper considerations, including which information users actually need, how complex financial data should be communicated, what interactions build trust, and whether the experience helps people make better decisions.</p><p>These questions require research, contextual understanding, and product judgment. They also require designers to consider how different users interpret information, how accessibility affects the experience, and how design decisions influence behavior over time. AI can accelerate the creation of potential solutions, but it cannot automatically establish whether those solutions are appropriate, valuable, or effective in a specific product context.</p><p>The distinction becomes especially important as AI-generated interfaces become more visually sophisticated. When almost anyone can produce an attractive dashboard or landing page, visual polish alone becomes a weaker indicator of product quality. The real differentiator is the reasoning behind the experience and the evidence that supports its design decisions.</p><h4>The Real Shift Is From AI Tools to Integrated Product Workflows</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-t01Zi6evOAEZotDDcn_Og.png" /><figcaption>Integrated Product Workflows</figcaption></figure><p>One of the most persistent limitations of traditional product development is fragmentation. Research, interface design, prototyping, engineering, and analytics often take place across separate tools and workflows. Moving between these environments introduces friction, slows collaboration, and creates opportunities for information to be lost or misinterpreted. Designers may spend significant time translating concepts into specifications, while developers must interpret design decisions that are not always fully documented.</p><p>AI presents an opportunity to reduce some of that fragmentation by connecting activities that have historically operated independently. Instead of treating AI as a tool that produces individual screens, product teams can explore how it supports the entire development lifecycle. Research findings can inform early concepts, concepts can evolve into functional prototypes, and prototypes can be tested and refined through faster feedback cycles. When these activities are connected, teams can spend less time transferring information between tools and more time evaluating whether their solutions are effective.</p><p>This changes the relationship between design and implementation. The traditional distinction between designing a product and building it becomes less rigid when designers can create functional experiences directly from their ideas. A product designer who understands interface architecture, component systems, and basic development principles can increasingly move from a design hypothesis to a working prototype without waiting for every stage of the traditional handoff process.</p><p>That does not eliminate the need for engineering expertise. Production systems still require careful attention to security, performance, maintainability, scalability, and technical reliability. However, AI-assisted development can make early exploration more practical and allow designers and engineers to collaborate around functional experiences rather than static representations.</p><p>Claude’s integration of design capabilities into Artifacts reflects this broader direction toward connected workflows. Although the decision alone does not establish whether Anthropic’s strategy will succeed, it illustrates how design generation can become part of a larger environment where users create, modify, and interact with different types of digital artifacts.</p><p>The real opportunity is not simply to generate interfaces faster. It is to reduce the distance between identifying a problem, exploring a solution, and validating its value. When teams can move through these stages more efficiently, they have more opportunities to challenge assumptions before committing substantial resources to development.</p><h4>Why Design Judgment Becomes More Valuable</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qHbJkUthEeqXs56qSVy4BA.png" /><figcaption>Human Judgment + AI</figcaption></figure><p>As AI makes interface generation increasingly accessible, the ability to produce screens becomes less distinctive. What becomes more valuable is the ability to evaluate, refine, and connect those screens to meaningful product outcomes. Designers are not simply responsible for making interfaces visually appealing; they are responsible for making decisions that influence how people understand, navigate, and experience digital products.</p><p>Design judgment extends far beyond visual aesthetics. It involves understanding how people perceive information, how interactions influence behavior, and how individual decisions contribute to a coherent product experience. It also requires recognizing when familiar patterns are appropriate and when a product needs a different approach because its users, context, or business model demand something more specific.</p><p>Visual quality remains important because typography, hierarchy, spacing, composition, and interaction details influence usability and trust. A well-designed interface can make complex information easier to understand and reduce the effort required to complete important tasks. However, strong visual execution must be supported by equally strong product reasoning, particularly when a design decision affects accessibility, user confidence, or business performance.</p><p>A beautifully designed onboarding experience may still fail if it introduces unnecessary steps or asks users for information before demonstrating value. A sophisticated analytics dashboard may create confusion if its information hierarchy does not reflect user priorities. A technically functional product may struggle with retention if its core value is not communicated effectively or if users cannot develop meaningful habits around its features.</p><p>These problems cannot be solved through visual generation alone because they depend on understanding the relationship between the product and the people using it. Designers need to interpret research findings, evaluate behavioral patterns, consider business constraints, and recognize the trade-offs involved in different solutions. They must understand when an experience is unnecessarily complex, when an interaction creates friction, and when a feature does not contribute meaningful value.</p><p>As AI accelerates execution, this ability to make informed decisions becomes an increasingly important differentiator. The designer’s contribution shifts from producing every visual element manually toward directing the quality, coherence, and effectiveness of the overall experience. The value lies not only in what gets created, but also in what gets questioned, refined, simplified, or deliberately left out.</p><h4>AI Changes Where Creativity Happens</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MI9wNPnhZGc5BYCCacs4cw.png" /><figcaption>Product Strategy &amp; Outcomes</figcaption></figure><p>The discussion around AI and creativity often assumes that automation and human expertise are competing forces. In practice, the relationship is more complex because creativity involves much more than producing visual artifacts. It also involves recognizing opportunities, reframing ambiguous problems, connecting seemingly unrelated ideas, and developing solutions that challenge established patterns.</p><p>AI can be valuable for repetitive production tasks, exploring alternative layouts, generating initial components, documenting design systems, and supporting implementation. By reducing the time required for these activities, designers can potentially devote more attention to research, experimentation, and strategic problem-solving. Instead of spending hours recreating familiar interface patterns, they can explore several directions and evaluate which one best supports the product’s goals.</p><p>However, the ability to generate many alternatives does not automatically lead to better design. Without a clear understanding of the problem, AI can simply produce more variations of the same conventional solution. A designer might generate ten different dashboard layouts, but if none addresses the user’s actual decision-making needs, the additional output creates activity rather than progress.</p><p>This is where creative judgment becomes essential. Designers must determine which ideas deserve further exploration, which patterns should be challenged, and which solutions are appropriate for a particular audience. They also need to recognize when an AI-generated result is visually convincing but conceptually weak, technically unrealistic, or inconsistent with the broader product experience.</p><p>The challenge is not to avoid automation, but to understand where it contributes value. Designers who use AI without critically evaluating its outputs risk producing generic experiences that look polished but lack meaningful differentiation. Conversely, designers who combine AI-assisted exploration with strong research, visual judgment, and strategic thinking can use the technology to expand their creative possibilities.</p><p>AI can increase the number of possibilities available to designers, but human expertise helps determine which possibilities are worth pursuing. This distinction matters because the purpose of design is not to maximize the number of outputs. It is to develop solutions that improve the relationship between people, technology, and the problems they are trying to solve.</p><h4>Product Strategy Is Becoming a Competitive Advantage</h4><p>As the boundaries between design and development evolve, the role of the product designer is expanding. Designers are increasingly able to contribute beyond interface creation by participating in product discovery, prioritization, experimentation, implementation, and performance evaluation. AI-assisted tools can support this broader involvement by making technical execution more accessible and reducing the effort required to test ideas.</p><p>However, faster execution does not guarantee better outcomes. A team can generate multiple prototypes in a single afternoon and still fail to identify the right problem. It can develop sophisticated features without understanding whether users need them, and it can release polished experiences without establishing how success will be measured. When production becomes easier, the risk of building unnecessary or poorly prioritized features can actually increase.</p><p>This is why product strategy becomes increasingly important. Effective product development requires clear priorities, evidence-based decisions, and a continuous understanding of how design choices influence user and business outcomes. Teams need to understand what they are trying to achieve, which assumptions need validation, and how they will determine whether a solution is successful.</p><p>For example, improving a SaaS onboarding experience should not begin with generating a more attractive welcome screen. It should begin with understanding where users encounter friction, how quickly they experience the product’s core value, and which behaviors indicate successful activation. Once these questions are understood, AI can help teams explore alternative flows, prototype solutions, and accelerate implementation.</p><p>The same principle applies to retention, conversion, accessibility, and product growth. AI becomes most valuable when it supports a clear objective rather than simply increasing the volume of work being produced. The ability to connect research insights, design decisions, and measurable outcomes is what transforms faster execution into meaningful product improvement.</p><p>For designers, technical fluency and AI capabilities are valuable because they expand what can be explored and delivered. But their impact depends on the strategic thinking guiding those capabilities. Understanding product metrics, business models, user behavior, and implementation constraints allows designers to contribute more effectively to decisions that shape the direction of a product.</p><p>The strongest product designers will not necessarily be those who generate the most interfaces or use the largest number of AI tools. They will be those who can connect user understanding, visual judgment, technical execution, and measurable business value. In an environment where execution is becoming increasingly accessible, knowing what to build and why it matters becomes a significant competitive advantage.</p><h4>The Future of Product Design Is About Better Decisions</h4><p>The transition of Claude Design into Artifacts should not be treated as proof that AI design has failed or that designers are immune to technological change. It is better understood as one example of how AI capabilities are being reorganized within broader product ecosystems. The long-term success of these approaches will depend on whether they genuinely improve workflows, reduce friction, and help people achieve better outcomes.</p><p>AI is changing the economics of execution by making prototypes faster to produce, implementation more accessible, and experimentation easier to initiate. These developments create significant opportunities for designers and product teams, but they also increase the importance of thoughtful decision-making. When the cost of creating something decreases, the ability to distinguish valuable ideas from unnecessary ones becomes even more important.</p><p>The central question is no longer simply whether AI can generate a good-looking interface. Increasingly, the question is whether teams can use these capabilities to identify meaningful problems, develop effective solutions, validate assumptions, and create measurable value. This requires a combination of creativity, research, product strategy, systems thinking, and technical understanding.</p><p>For me, the future of product design is not about choosing between human creativity and artificial intelligence. It is about combining both in ways that improve the quality of the products we build. Designers who approach AI as a strategic capability rather than a shortcut for generating screens will be better positioned to influence how digital products are conceived, developed, and improved.</p><p><strong>AI accelerates execution. Human judgment gives it direction. Product strategy connects both to meaningful outcomes.</strong></p><p>That is where the next evolution of product design begins.</p><h4>References</h4><p><a href="https://claude.ai/">Anthropic Claude</a></p><p><a href="https://support.claude.com/en/articles/17440474-move-from-standalone-claude-design-to-claude">Claude Design Migration — Anthropic Help Center</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a13d221b3b9f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What Synthetic Users Leave Out]]></title>
            <link>https://medium.com/space-ux-design/what-synthetic-users-leave-out-9a6084c61eb6?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/9a6084c61eb6</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[ux-research]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Fri, 06 Nov 2026 13:18:41 GMT</pubDate>
            <atom:updated>2026-11-06T13:40:48.871Z</atom:updated>
            <content:encoded><![CDATA[<p><em>AI can help us prepare for research, but it cannot yet replace what we learn from people.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8CPbJ3HzZI275Xdd-UImMg.png" /></figure><p>A recent Nielsen Norman Group study raises an important question about how we evaluate synthetic respondents. If AI-generated answers lead us to the same broad conclusion as human research, does that mean they are a reliable substitute?</p><p>In this comparison, researchers asked 120 synthetic respondents a question previously answered by 604 people about what to call UX work. UX-related terminology dominated both datasets, but the range of answers differed considerably. Human respondents used more than 50 distinct labels, while synthetic respondents produced about a dozen. The overall conclusion was similar, yet the synthetic answers captured much less variation. <a href="https://www.nngroup.com/articles/synthetic-users-vs-human-respondents/">Read the NN/G study</a>.</p><p>For a product team, that difference deserves attention. We often begin research with a reasonable explanation of what people need and how they might behave. What we do not know is where that explanation falls short. When generated responses cluster around familiar answers, they may make our understanding feel more complete without helping us discover what is missing.</p><h4>What unusual answers can reveal</h4><p>An unexpected answer can be the beginning of a useful investigation. Someone might interpret a familiar word differently, describe a task the team has overlooked, or explain a workaround that has become part of their routine. Another person might struggle with something that most participants find straightforward.</p><p>These differences do not all deserve the same attention. An unusual response might reveal a serious barrier, or it might have little relevance to the decision at hand. Research gives us a way to explore the circumstances behind an answer before deciding what it means.</p><p>We can ask someone to explain what happened, show us how they completed a task, or describe the last time they encountered a problem. As the conversation develops, we may discover that our original question was too narrow or that we used language the participant understood differently.</p><p>Consider a hypothetical participant who describes a checkout process as easy but repeatedly switches between the payment page and their banking app. That behavior gives us a reason to ask more questions. They might be checking their balance, looking for payment details, or confirming that a previous attempt worked. Each explanation points toward a different issue, and observation alone would not tell us which one is correct.</p><p>This is part of what research contributes: an opportunity to connect an answer with the situation that produced it. We can follow up on uncertainty instead of immediately turning it into a finding.</p><h4>Research helps us question the problem itself</h4><p>Product teams usually enter a study with decisions already taking shape. There may be a feature to prioritize, a concept to evaluate, or a journey that needs improvement. Those practical goals help focus the work, but they can also limit what the team is prepared to notice.</p><p>Contact with people creates opportunities to reconsider that focus. We may learn that the task we want to simplify happens rarely, that a proposed feature duplicates an existing workaround, or that the most frustrating part of an experience occurs outside our product.</p><p>For example, a team might assume that people abandon a form because it takes too long. A conversation could reveal that some people stop because they do not have the required information or do not understand why it is being requested. Shortening the form might help, but it would not necessarily address those concerns.</p><p>Human research has its own limitations. Recruitment affects whose experiences we hear, questions influence responses, and researchers bring assumptions to interpretation. Speaking to people does not automatically produce an accurate understanding. It does, however, allow people to introduce experiences and priorities that the team did not know to ask about.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gJxrK1gULsoPfmeLUif4Ug.png" /></figure><h4>Why convincing answers are not enough</h4><p>Synthetic respondents can produce detailed, coherent explanations. The challenge is establishing whether those explanations reflect the audience we intend to serve.</p><p>A generated response might describe a concern that sounds entirely reasonable. Before using it to guide a decision, we still need to know whether people actually encounter that concern, under what circumstances it matters, and what they do when it happens. A plausible explanation cannot answer those questions by itself.</p><p>Asking for more varied responses does not settle the issue either. We would still need evidence that the additional perspectives reflect real experiences rather than possibilities introduced to satisfy the prompt.</p><p>The NN/G study should also be read within its scope. It compared one question using one platform, and the human and synthetic groups differed in composition and response format. Those limitations matter when interpreting the findings. <a href="https://www.nngroup.com/articles/synthetic-users-vs-human-respondents/">See the study’s methods and limitations</a>.</p><p>For me, the practical conclusion is that synthetic respondents are not sufficient, for now, to establish what a particular audience needs, how it behaves, or whether a concept works for it. They may give us useful possibilities to investigate, but those possibilities need evidence before they carry the weight of a research finding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VCgX18RzsNZu9gyxmjmLcw.png" /></figure><h4>Where synthetic respondents can help</h4><p>I would use synthetic respondents during preparation, when a team is exploring what to ask and which assumptions to examine. They can suggest possible objections, offer alternative hypotheses, or help identify gaps in an interview guide.</p><p>Suppose a team is preparing to research a subscription service. Generated responses might prompt questions about cancellation, payment timing, or uncertainty about ongoing costs. Those suggestions could improve the discussion guide, provided the team leaves room for participants to explain what matters to them in their own terms.</p><p>This use also requires care. If we take every generated concern into an interview as something to confirm, we risk leading participants toward the model’s framing. The preparation should help us become more curious and attentive, while keeping our conclusions open.</p><h4>Keep the source visible</h4><p>The distinction between a generated idea and a research finding needs to remain clear as information moves through a team. A speculative response can lose its context when it is copied into a presentation or summarized in a planning document.</p><p>Over time, a possible concern can become something that “users told us.” A generated preference can become a reason to prioritize a feature, even though nobody from the intended audience expressed it.</p><p>Research notes and decision documents should make the source visible. Readers should be able to understand whether a statement came from a participant, an observed behavior, an interpretation, or a generated hypothesis. They also need enough context to judge what that evidence can support.</p><p>That clarity makes decisions easier to review. If a team chooses to act on an assumption, it can acknowledge the uncertainty and plan how to investigate it.</p><h4>What are we trying to save time on?</h4><p>There is a useful role for AI in making research preparation more efficient. It can help teams explore possibilities and develop questions before meeting participants. The value of that work depends on whether it improves what happens next.</p><p>If the time saved allows us to listen more carefully, investigate uncertainty, or include perspectives we might otherwise miss, it can strengthen the research process. If it removes contact with the people whose behavior we need to understand, we may reach a confident conclusion sooner while leaving important questions unanswered.</p><p>Research helps us discover where our understanding needs to change. Sometimes that happens through a surprising answer. Sometimes it happens through a hesitation, an improvised workaround, or an explanation that makes us reconsider what we thought the problem was.</p><p>Synthetic respondents can help us articulate what we expect to hear. Research with people gives us an opportunity to learn why those expectations may be incomplete.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9a6084c61eb6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/what-synthetic-users-leave-out-9a6084c61eb6">What Synthetic Users Leave Out</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Haiku 5.5 vs. Opus 5.5: Why I Choose Opus for Complex UX and Product Design]]></title>
            <link>https://medium.com/space-ux-design/haiku-5-5-vs-opus-5-5-why-i-choose-opus-for-complex-ux-and-product-design-2a75ddef5d3a?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2a75ddef5d3a</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[claude-ai]]></category>
            <category><![CDATA[product-design]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Tue, 13 Oct 2026 12:22:31 GMT</pubDate>
            <atom:updated>2026-11-06T13:26:49.969Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Why I prioritize contextual understanding, thoughtful reasoning, and fewer revisions over faster AI responses in my daily design workflow.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NXbS1YolQmIe-G6D43Hi6A.png" /></figure><p>As a Senior Product Designer, AI has become a natural part of how I approach my everyday work. I use it to explore ideas, improve user experiences, build interfaces, and manage design systems in Figma. Over time, I’ve experimented with different models to understand which ones are genuinely useful for the problems I deal with.</p><p>Recently, I’ve been looking at Haiku 5.5 Medium and Opus 5.5 Medium in Claude. Both offer Medium thinking effort, but I don’t necessarily see them as interchangeable. For the type of work I do, I prefer Opus because I value contextual understanding and thoughtful reasoning more than getting the fastest possible response.</p><p>That doesn’t mean I think Haiku is a bad choice. I believe lightweight models have an important place in everyday workflows. However, when I’m working on something involving multiple screens, existing components, or complex product decisions, I would rather use a model that can help me think through the problem more carefully.</p><p>This isn’t a controlled performance benchmark. It’s my perspective on how I choose AI models for different design tasks and why I’ve become more interested in the quality of the overall workflow than the speed of an individual response.</p><h4>1. Faster Responses Don’t Always Mean Better Productivity</h4><p>One thing I’ve noticed with AI-assisted design is that generating something quickly doesn’t always mean finishing the task quickly. I might ask AI to improve a navigation component, and within a short time, it produces something that looks good. But when I review the result, I sometimes discover that it hasn’t considered the existing design system, interaction patterns, or how the changes might affect other screens.</p><p>Then I need to explain the problem again, correct inconsistencies, and go through another round of revisions. What initially seemed like a quick task can become a much longer process.</p><p>This has changed how I think about productivity. I don’t really care whether AI responds in a few seconds or takes a little longer if the final result still requires significant corrections. What matters more to me is how much time and effort I spend getting to something I can actually use.</p><p>When I’m working on a complex product, I would rather spend a little more time getting a thoughtful response than spend additional time fixing something that was generated quickly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bG6ohnrSpj32JDgmZu5MDA.png" /><figcaption><em>Faster initial responses can still lead to longer workflows when multiple revisions are required. Conceptual illustration, not benchmark data.</em></figcaption></figure><h4>2. Understanding Product Context Matters More Than Generating Interfaces</h4><p>I think there’s an important difference between generating an interface and understanding a product problem. AI has become very good at producing visually polished designs, but a clean interface doesn’t necessarily mean the user experience has improved.</p><p>For example, if I’m working on a news application and users are having difficulty navigating between articles, the solution might not be to introduce new components or redesign the entire page. Sometimes a small change to the existing navigation can make the experience much better.</p><p>This is where I want AI to do more than follow instructions. I want it to consider why a problem exists, how users interact with the product, and whether the proposed solution actually improves the experience.</p><p>I’ve also noticed that even simple design instructions can be interpreted in different ways. If I ask AI to make an interface more minimal, it might remove elements or reduce the amount of information displayed. But that isn’t always what I mean. Sometimes the problem is the information hierarchy, inconsistent spacing, or too many competing actions.</p><p>As designers, we know that making something minimal isn’t just about removing things. It’s about making the experience easier to understand while keeping the information and functionality users actually need.</p><p>That’s the kind of reasoning I value when working with AI.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iLaibNrWxFwOYm0JgJ742Q.png" /><figcaption><em>Good product decisions depend on understanding the relationships between user needs, behavior, accessibility, design systems, and business goals.</em></figcaption></figure><p>Another reason I prefer Opus for complex tasks is how I approach design systems. When working in Figma, I’m often dealing with components that are connected to multiple screens and different states. A small change to one component can affect the consistency of the entire product.</p><p>In these situations, I don’t want AI to create unnecessary components or introduce changes that don’t follow the existing structure. I want it to understand what already exists and work within those constraints.</p><p>Of course, no model can understand the entire product automatically. I still need to provide context, explain requirements, and review the output. But I think choosing a model with stronger reasoning capabilities can be valuable when the work involves several connected decisions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8DC0O_s1eQK8OcQYZULmWQ.png" /><figcaption><em>Following a design instruction literally isn’t always the same as understanding the intention behind it.</em></figcaption></figure><h4>3. Choosing Between Haiku and Opus Depends on the Task</h4><p>When I think about Haiku 5.5 Medium and Opus 5.5 Medium, I don’t believe one model needs to handle everything.</p><p>There are plenty of tasks where I don’t need extensive reasoning. If I’m rewriting a short message, improving microcopy, or generating a few initial ideas, a lightweight model like Haiku can make sense. There’s no reason to use a more resource-intensive model when a simpler one can produce a useful result.</p><p>For me, choosing between models is really about understanding the complexity of the task. I would rather use Haiku for straightforward work and Opus when I need to explore more complicated UX problems or product decisions.</p><p>I also think it’s important to recognize that using the same Medium thinking-effort setting doesn’t make the two models equivalent. Their underlying capabilities can differ, and the same setting doesn’t guarantee identical reasoning behavior or performance.</p><p>When I’m working on a design system, I need to think about consistency, dependencies, component behavior, and how changes affect the rest of the product. When I’m evaluating a UX problem, I want to explore the reasoning behind different solutions rather than immediately generating a new interface.</p><p>These are the situations where I would rather work with Opus.</p><p>That doesn’t mean every response needs to be longer or more complicated. In fact, I often prefer concise recommendations. What matters to me is whether the model has considered the problem carefully enough to provide a useful answer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VF_m3y40HDI6cuvOK37Ebw.png" /><figcaption><em>My preferred approach is to match the AI model to the complexity of the task rather than use the same model for everything.</em></figcaption></figure><h4>4. The Hidden Cost of AI Is the Work We Have to Redo</h4><p>Something I’ve been thinking about recently is how much time we actually spend correcting AI-generated work.</p><p>We often focus on how quickly AI can build an interface, generate code, or complete a design task. But we don’t always consider how much work happens afterward.</p><p>If AI generates a design in a few minutes but I spend another hour correcting components, fixing inconsistencies, and reviewing interactions, was the process actually efficient?</p><p>I think this is an important question, especially as AI becomes more integrated into tools like Figma and other product development environments.</p><p>Imagine asking AI to update a component library. The model makes the requested changes, and everything initially looks fine. Then I notice that some components have been duplicated, a few variants aren’t consistent, and certain screens are still using the previous version.</p><p>Now I need to identify what went wrong, explain the problem, and ask for another round of changes.</p><p>Sometimes the corrections are small. Other times, they become more complicated than the original task.</p><p>This is why I think we should pay more attention to the complete workflow rather than the initial generation time.</p><p>For me, the real value of AI isn’t simply how much it can generate. It’s how much unnecessary work it helps me avoid.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gTuEefcXBxjv-6kHIEqQkA.png" /><figcaption><em>The time spent reviewing and correcting AI-generated work is an important part of evaluating productivity.</em></figcaption></figure><p>I also don’t think AI should replace the designer’s judgment. Even when a model provides a convincing recommendation, it doesn’t necessarily understand every business constraint, user behavior, or previous decision behind the product.</p><p>AI can help explore alternatives and identify potential issues, but I still believe designers need to make the final decisions. We understand the product context, evaluate the trade-offs, and take responsibility for the experience we’re creating.</p><p>This is why I see AI as a collaborator rather than something that should make every decision for me. I want it to help me work more efficiently, question my assumptions, and explore better solutions without losing control over the design process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I4wOYTEZ3UW4M8dhKjK1GA.png" /><figcaption><em>AI can support product thinking and help evaluate alternatives, but the responsibility for the final decision remains with the designer.</em></figcaption></figure><h4>5. Why I Prefer Opus 5.5 Medium for My Design Workflow</h4><p>When I think about Haiku 5.5 Medium and Opus 5.5 Medium, my preference comes down to what I value most in my daily work.</p><p>I haven’t conducted a controlled benchmark using identical tasks, so I wouldn’t claim that Opus performs better in every situation. However, for the complex product design work I want AI to support, Opus is the model I would choose.</p><p>I’m also interested in seeing how lightweight models evolve. As smaller models become more capable, the differences between them may become less noticeable for many everyday tasks. That could make AI-assisted workflows faster, more affordable, and more accessible.</p><p>For now, though, I prefer to prioritize the quality of the result over the speed of the first response.</p><p>I think we’re moving toward a point where the most important question won’t be how quickly AI can generate something, but how well it understands what we’re trying to accomplish.</p><p>As designers, we don’t necessarily need AI to create more interfaces or suggest more features. We need tools that help us understand problems, evaluate decisions, and build better experiences.</p><p>That’s ultimately why I prefer Opus 5.5 Medium for my work.</p><p><strong>I don’t need AI to design more. I need it to help me design better.</strong></p><p>I’m curious how other designers approach this. Do you prefer faster models for everyday work, or do you find that more capable models are worth using when dealing with complex design problems?</p><p><em>Fatma Doğan Seçkin is a Senior Product Designer focused on AI-powered product experiences, UX strategy, and product design.</em></p><p><a href="https://fatmadoganseckin.com/">fatmadoganseckin.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2a75ddef5d3a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/haiku-5-5-vs-opus-5-5-why-i-choose-opus-for-complex-ux-and-product-design-2a75ddef5d3a">Haiku 5.5 vs. Opus 5.5: Why I Choose Opus for Complex UX and Product Design</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beyond the Chatbox: Designing UX for Autonomous AI Agents]]></title>
            <link>https://medium.com/space-ux-design/beyond-the-chatbox-designing-ux-for-autonomous-ai-agents-381ddd6d6c06?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/381ddd6d6c06</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[chatgpt]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Mon, 12 Oct 2026 07:01:01 GMT</pubDate>
            <atom:updated>2026-11-06T13:27:06.219Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*J_J-k0__B7Fs3j7hwFyC8g.png" /></figure><p>Why the future of AI product design is shifting from conversations to delegation, control, and trust.</p><p>For the past few years, much of AI product design has revolved around one familiar interface: the chatbox. We type a prompt, receive a response, and continue the conversation until we get what we need. It is simple, familiar, and has made artificial intelligence accessible to millions of people.</p><p>But as AI becomes more capable, the way we interact with it is starting to change.</p><p>We’re moving beyond systems that simply answer questions. Autonomous AI agents can now plan tasks, work across different tools, retrieve information, and execute workflows with varying levels of human involvement. Instead of guiding every individual action, users can increasingly define what they want to achieve and let AI handle parts of the process.</p><p>This creates an interesting challenge for product designers. If AI no longer needs constant instructions, how do we design experiences that keep people informed, confident, and in control?</p><p>I believe this is one of the most important questions emerging in AI product design. The next generation of AI interfaces will need to support more than conversations. They’ll need to help people understand what their agents are doing, why certain decisions are being made, and when human involvement is necessary.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7IcTNai84fmRqpTWgvpZzw.png" /></figure><h3>1. From Prompting to Delegating</h3><p>Think about how we typically interact with AI today. We ask a question, review the answer, and decide what to do next. If the response isn’t quite right, we adjust the prompt and try again. The user remains responsible for moving the process forward.</p><p>Autonomous agents introduce a different experience. Instead of asking AI to complete one small task at a time, we can give it a broader objective and allow it to coordinate the steps needed to reach that goal.</p><p>Imagine a product manager preparing for a quarterly product review. Traditionally, they might need to gather analytics from several platforms, identify changes in user behavior, summarize customer feedback, and prepare a presentation. AI can already help with many of these activities, but the user often needs to coordinate the work manually.</p><p>With an autonomous agent connected to the appropriate tools, much of that coordination could happen within a single workflow. The product manager could define the objective, specify the information sources, and review the results as the agent progresses.</p><p>The important difference isn’t simply that AI can perform more tasks. It’s that the relationship between the user and the product begins to change.</p><p>Instead of focusing on individual instructions, users start thinking about outcomes, expectations, and responsibilities. They need to understand what the agent can do independently, what requires approval, and how to intervene when something doesn’t go as expected.</p><p>From a UX perspective, this changes how we approach interaction design. A well-designed experience shouldn’t require users to constantly monitor an agent, but it also shouldn’t leave them wondering what’s happening behind the scenes.</p><p>The challenge is finding the right balance between autonomy and control.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FkIjOLpyvKW6ualVBtqNbA.png" /></figure><h3>2. AI Agents Need More Than Conversational Interfaces</h3><p>Chat interfaces work particularly well for immediate interactions. When someone wants to summarize an article, generate ideas, or ask a question, a conversational interface feels natural.</p><p>However, the same approach becomes less effective when an AI system is working on a complex task over an extended period.</p><p>Imagine asking an agent to analyze customer feedback from multiple sources. The agent might retrieve survey responses, identify recurring complaints, compare findings with product analytics, and prepare recommendations for the product team.</p><p>If every operation appeared as a separate chat message, the conversation could quickly become difficult to follow. Users would have to scroll through technical updates just to understand whether the task was progressing as expected.</p><p>This is where I think AI product design needs to evolve.</p><p>Rather than treating the chatbox as the center of every interaction, we should start considering interfaces that communicate progress, responsibilities, and decisions more effectively.</p><p>An activity timeline, for example, could show what an agent has completed and what it’s currently working on. Users wouldn’t need to inspect every technical step, but they could access more information whenever necessary.</p><p>Permissions are another important part of the experience. If an agent needs access to external applications or sensitive information, users should understand what they’re authorizing. These controls should be presented clearly and in context, rather than hidden inside complicated settings.</p><p>Approval mechanisms also become essential. Not every action needs confirmation, but publishing content, sending external communications, or making significant changes to data may require explicit permission.</p><p>At the same time, users need to understand why an agent made a particular recommendation. This doesn’t mean displaying every internal calculation. It means providing meaningful explanations, relevant evidence, and enough context to evaluate the result.</p><p>And perhaps most importantly, users need ways to recover when things go wrong. They should be able to pause a workflow, correct an instruction, retry a failed operation, or reverse an action when the system supports it.</p><p>These are not entirely new UX principles. Visibility, user control, error prevention, and recovery have been part of interaction design for decades.</p><p>What’s changing is the context in which we apply them.</p><p>As AI systems become more autonomous, these principles become central to the experience rather than supporting features.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*A0whKbzavguuWLQ7gmTjOw.png" /></figure><h3>3. Trust Is Becoming a Core UX Challenge</h3><p>Trust has always been important in digital products. People expect applications to behave consistently, communicate clearly, and protect their information.</p><p>But autonomous AI introduces a new dimension.</p><p>When we use a traditional application, we usually initiate the actions ourselves. We click a button, submit a form, or confirm a transaction. Even when the interface isn’t perfect, we generally understand that our actions are driving the process.</p><p>With autonomous agents, some decisions and operations may happen without continuous human involvement.</p><p>That raises a different set of expectations.</p><p>Users need to know what the agent is allowed to do, when it will ask for permission, and how they can verify the results. They also need confidence that unexpected situations won’t lead to actions they never intended to authorize.</p><p>Consider the difference between an agent organizing internal documents and one sending a payment.</p><p>Organizing documents might be a relatively low-risk activity, especially if changes can be reversed. Sending money involves greater consequences and should require stronger safeguards.</p><p>Both tasks involve automation, but they shouldn’t necessarily follow the same interaction model.</p><p>This is why I believe autonomy should be designed around the consequences of an action rather than treated as a single feature that can simply be turned on or off.</p><p>For some tasks, the agent might only suggest actions. For others, it could execute approved workflows independently while keeping the user informed. More consequential operations might require explicit confirmation before proceeding.</p><p>The goal isn’t to make AI as autonomous as possible. It’s to make automation appropriate for the situation.</p><p>An agent that asks for approval at every step can become frustrating. But an agent that operates without meaningful transparency can quickly undermine trust.</p><p>Good UX helps people understand where those boundaries exist and gives them confidence that they can maintain control when it matters.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZRyGd2347Kkfb-TO0P4nAA.png" /></figure><h3>4. Human-in-the-Loop Is an Interaction Design Principle</h3><p>Human-in-the-loop is often discussed in the context of AI development and system oversight, but I think it deserves more attention as an interaction design principle.</p><p>The question isn’t simply whether a human should be involved. It’s about deciding when their involvement adds value and how to make that interaction as clear as possible.</p><p>Imagine an autonomous agent preparing a marketing campaign. It could research competitors, organize ideas, generate content, and prepare a publishing schedule without requiring constant supervision.</p><p>However, publishing that campaign is a different kind of action.</p><p>Before proceeding, the agent could present a concise review screen explaining what it intends to publish, which channels will be affected, and what content has been prepared.</p><p>The user could then approve the campaign, request changes, or reject the action.</p><p>This approach allows the agent to handle repetitive work while keeping meaningful decisions with the user.</p><p>It also demonstrates why progressive disclosure is so valuable in agent-based experiences.</p><p>Autonomous workflows can produce a significant amount of information, but users rarely need to see everything at once. Most of the time, they want to know whether the task is progressing, whether something requires their attention, and whether the outcome meets their expectations.</p><p>A well-designed interface can present this information at a high level while allowing users to explore the details when needed.</p><p>For example, a dashboard might initially show that an agent has completed its analysis and is waiting for approval. If the user wants to understand how the recommendations were generated, they could expand the activity history or review the supporting evidence.</p><p>This creates a more manageable experience without sacrificing transparency.</p><p>The purpose of human oversight shouldn’t be to make users supervise every operation. It should be to involve them at the moments where their judgment is most valuable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g5VChaDbQAOPeUNwU3nVsA.png" /></figure><h3>5. Designing Agent-Native Products</h3><p>Most SaaS products are built around direct interaction. Users navigate dashboards, complete forms, update records, and move between different sections to accomplish their goals.</p><p>Autonomous agents introduce the possibility of a different model.</p><p>Instead of navigating through several interfaces to complete a workflow, users may be able to describe the outcome they want and let an agent coordinate the necessary steps.</p><p>That doesn’t mean traditional interfaces will disappear. In fact, I think dashboards, navigation systems, and familiar interaction patterns will remain important.</p><p>Their purpose, however, may begin to change.</p><p>Consider a project management platform. Today, much of the experience focuses on creating tasks, assigning responsibilities, updating statuses, and monitoring deadlines.</p><p>An agent-native version of the same product could allow users to define project goals and delegate certain operational activities to AI.</p><p>The interface might then focus more heavily on showing active workflows, highlighting decisions that require attention, and explaining what has changed since the user’s last visit.</p><p>Rather than manually updating every task, a project manager might review an agent’s proposed changes and approve the next steps.</p><p>This introduces a shift from task execution toward task supervision.</p><p>It also creates new opportunities for product designers.</p><p>We can explore how to make agent activity understandable, how to organize approval requests, how to communicate uncertainty, and how to design interfaces that support collaboration between people and autonomous systems.</p><p>For me, the most interesting part is that these experiences don’t necessarily require entirely new visual patterns.</p><p>Many familiar UX concepts can still work. The challenge is adapting them to systems that are capable of acting independently.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g5VChaDbQAOPeUNwU3nVsA.png" /></figure><h3>6. Accessibility Cannot Be an Afterthought</h3><p>As AI interfaces become more sophisticated, accessibility needs to remain a fundamental part of the design process.</p><p>Agent-based products often rely on dynamic information, including progress indicators, activity timelines, notifications, and interactive approval controls.</p><p>If these elements aren’t designed carefully, they can create significant barriers.</p><p>For example, an agent’s status shouldn’t be communicated through color alone. Users should also receive clear text labels explaining whether a task is running, completed, paused, or waiting for approval.</p><p>Similarly, approval actions should be accessible through keyboard navigation and clearly labeled for assistive technologies.</p><p>Dynamic updates need particular attention. If an agent continuously changes its status, screen reader users shouldn’t be overwhelmed by unnecessary announcements. Important changes should be communicated appropriately, while less critical information can remain available on demand.</p><p>Readable typography, sufficient contrast, predictable navigation, and clear interaction states are equally important.</p><p>WCAG 2.2 AA provides a valuable foundation for these decisions, but accessibility also requires thinking about the overall experience.</p><p>Can users understand what an agent is doing without constantly monitoring the interface? Can they identify when their attention is required? Can they pause or correct a workflow without navigating through several complicated screens?</p><p>These questions matter because autonomous AI should reduce unnecessary effort, not introduce new barriers.</p><p>I believe accessibility and agent UX are closely connected through a shared objective: making complex systems understandable and controllable for the people using them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fkOyMWDgNA_SjGRSW_As_w.png" /></figure><h3>7. How Should We Measure Agent UX?</h3><p>As AI products become more autonomous, we also need to reconsider how we evaluate their experiences.</p><p>Traditional UX metrics such as task completion, time on task, conversion, and usability remain relevant. But they don’t necessarily tell us whether an autonomous system is working well from the user’s perspective.</p><p>An agent might complete a task quickly while making decisions the user doesn’t understand or trust.</p><p>Similarly, a workflow might appear efficient because it requires very little human intervention, even though users have difficulty identifying when something goes wrong.</p><p>This suggests that agent-based experiences need a broader approach to measurement.</p><p>Task success remains important because the agent ultimately needs to achieve the intended outcome. But we should also examine how frequently users intervene, how efficiently they handle approval requests, and whether they can recover when errors occur.</p><p>Permission changes can also provide useful signals. If users repeatedly adjust what an agent is allowed to do, that behavior may reveal uncertainty about the system’s capabilities or a mismatch between its default settings and user expectations.</p><p>Trust and confidence deserve attention as well.</p><p>These qualities are more difficult to measure through behavioral analytics alone, which makes qualitative research particularly valuable. Interviews, usability testing, and longitudinal studies can help us understand whether users feel comfortable delegating responsibilities over time.</p><p>However, I would be careful about interpreting any single metric in isolation.</p><p>A low intervention rate doesn’t automatically indicate a successful experience. It might mean the agent is performing reliably, but it could also mean users don’t know how to intervene.</p><p>Likewise, faster completion isn’t always better if it comes at the expense of accuracy, transparency, or control.</p><p>I think the real opportunity is to evaluate autonomous AI products through both operational performance and human experience.</p><p>We shouldn’t only ask whether the agent completed the task.</p><p>We should also ask whether the user understood what happened, trusted the result, and felt capable of taking control when necessary.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IyvfbQR0XTSHLaRElZZzmQ.png" /></figure><h3>8. What This Means for Product Designers</h3><p>Working with AI-assisted design and development tools has changed how I approach certain parts of the product design process.</p><p>Tasks that previously required considerable time, such as exploring interface concepts, generating prototypes, or testing different implementation approaches, can now happen much faster.</p><p>This creates opportunities to experiment, validate ideas, and move from concepts to working experiences with less friction.</p><p>But I’ve also noticed that faster execution doesn’t automatically lead to better design decisions.</p><p>AI can generate an interface that looks polished and technically convincing while missing important details about the user’s context, expectations, or actual problem.</p><p>Sometimes it introduces unnecessary complexity. Other times, it produces solutions that appear reasonable but don’t align with the product’s goals.</p><p>That’s why I see AI as a powerful part of the design process rather than a replacement for product thinking.</p><p>As autonomous agents become more common, I believe this distinction will become even more important.</p><p>Designers will need to think beyond individual screens and consider how entire systems behave.</p><p>We’ll need to define appropriate boundaries, establish meaningful feedback mechanisms, anticipate failure scenarios, and ensure users can understand and control increasingly complex workflows.</p><p>This requires a combination of interaction design, product strategy, systems thinking, and research.</p><p>The questions we ask will evolve, too.</p><p>Instead of focusing only on how to make an interface easier to use, we’ll also need to understand how people develop trust in autonomous systems, how much control they expect, and when they want to participate in decisions.</p><p>For me, these are some of the most interesting challenges in the next phase of AI product design.</p><h3>Final Thoughts: The Future of AI UX Is About Control, Not Just Automation</h3><p>The chatbox played an important role in making artificial intelligence accessible. It gave people a familiar way to interact with technology that once felt complicated or inaccessible.</p><p>Autonomous agents are taking that experience in a new direction.</p><p>As AI becomes capable of planning and executing more complex workflows, the relationship between users and digital products will continue to evolve.</p><p>But greater autonomy doesn’t automatically create a better experience.</p><p>People still need to understand what’s happening, make informed decisions, and feel confident that they can intervene when necessary.</p><p>That’s where thoughtful product design becomes essential.</p><p>I don’t think the future of AI UX is about replacing every interaction with automation or removing people from the process.</p><p>It’s about understanding which responsibilities can be delegated, which decisions should remain with users, and how technology can support that relationship in a meaningful way.</p><p>As designers, we have an opportunity to shape these experiences before their interaction patterns become established conventions.</p><p>And perhaps that’s the most interesting part of this shift.</p><p><strong>We’re no longer designing only how people interact with AI. We’re increasingly designing how people and AI work together.</strong></p><p>That future extends far beyond the chatbox.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=381ddd6d6c06" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/beyond-the-chatbox-designing-ux-for-autonomous-ai-agents-381ddd6d6c06">Beyond the Chatbox: Designing UX for Autonomous AI Agents</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Accessibility Is Not a Feature, It Is the Foundation of Design]]></title>
            <link>https://medium.com/space-ux-design/accessibility-is-not-a-feature-it-is-the-foundation-of-design-ea4e1f617a13?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/ea4e1f617a13</guid>
            <category><![CDATA[accessibility]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[inclusive-design]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Fri, 07 Aug 2026 10:07:42 GMT</pubDate>
            <atom:updated>2026-11-06T13:27:23.336Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NMmG5ZOM638554KQg1PNDw.png" /><figcaption>Cover — Accessibility · Design: Fatma Doğan Seçkin</figcaption></figure><p>As a Senior Product Designer, I have been working on accessibility for a long time, and I see WCAG standards, color contrast, screen readers, and keyboard navigation as natural parts of the product development process. Over time, however, I realized something important: sometimes, knowing the standards is not enough to truly understand a subject. You need to listen to the people who actually live through those experiences.</p><p>Today, I attended <strong>Accessibility Roundtable #1: Integrating Accessibility into Your Workflows</strong>, organized by <a href="https://www.linkedin.com/company/binclusive/"><strong>Binclusive</strong></a><strong>, T</strong><a href="https://www.linkedin.com/company/turkishtechnology/"><strong>urkish Technology</strong></a><strong>, and </strong><a href="https://www.linkedin.com/school/brick-institute/"><strong>Brick Institute</strong></a>, for the first time. First of all, I would like to thank everyone who made this valuable event possible. My only wish is that events like this are not limited to in-person participation and can also be organized online. Every discussion about accessibility deserves to reach more designers, developers, product managers, and decision-makers.</p><p>Throughout the event, we had productive discussions about the future of accessibility in Turkey, how companies should prepare for this transformation, and how digital products can become more inclusive.</p><p>At the center of these discussions was an important regulation. <strong>Presidential Circular №2025/10</strong>, published in the Official Gazette <strong>№32933</strong> on <strong>June 21, 2025</strong>, under the title <em>“Accessibility of Websites and Mobile Applications,”</em> made accessibility a legal requirement for digital products and services used by people with disabilities and older adults in Turkey. The regulation concerns not only public institutions but also the private sector. Accessibility is therefore no longer simply a good user experience practice; it is also a legal responsibility that needs to be considered throughout the product development process.</p><p>Yet by the end of the day, what stayed with me was neither the new regulation nor the technical requirements.</p><p>What affected me most was hearing visually impaired participants describe the challenges they face in everyday life through their own experiences.</p><p>At that moment, I realized that we still often treat accessibility as a checklist. We check contrast ratios, add alternative text, test screen reader compatibility, and make sure keyboard navigation works. Once those tasks are complete, we tend to assume that our job is done.</p><p>But accessibility is much more than that.</p><p>To me, accessibility means enabling people to live independently in both the digital and physical world. Being able to use a banking application without relying on someone else, purchase the product you want online independently, or use public transportation on your own are not additional features. They are some of the most fundamental responsibilities of a well-designed product.</p><p>Another thing that impressed me during the event was that a significant part of the team working in accessibility consisted of visually impaired people themselves. It reminded me of one of the most fundamental principles in UX:</p><p><strong>Design with users, not just for users.</strong></p><p>People who experience a problem firsthand can often make the most valuable contributions to solving it. That is why I believe people with disabilities should not only be involved at the final stage of product testing. They should also be a natural part of research, design, and product development processes. Truly inclusive products can only emerge from this kind of approach.</p><p>The everyday examples shared during the event also showed how many barriers most of us simply do not notice. A visually impaired person may stand in front of a supermarket shelf and appear to have found the product they want. But if they cannot read the information on the packaging, they may have no way of knowing whether they have actually chosen the right item.</p><p>The same thing happens in the digital world. When an e-commerce website does not provide alternative text for product images, when campaigns are communicated only through visuals, or when buttons are not properly announced by screen readers, tasks that take many of us only a few seconds can become serious barriers for someone else.</p><p>No one is asking for special treatment.</p><p>People simply want to live independently, move freely, and purchase the products they want just like everyone else.</p><p>During the event, one of the organizers shared an experience from a visit to Sweden. They had seen a staff member waiting at a bus stop to meet and assist a visually impaired passenger. That example stayed with me. My first assumption was that this might have been enabled by Bluetooth-based technology.</p><p>When I got home, I asked myself:</p><p><strong>How can a system know that a visually impaired passenger needs assistance before they even arrive at the stop?</strong></p><p>As I started researching this question, I realized that accessibility is not limited to mobile applications or websites. Cities themselves are also designed user experiences.</p><p>During my research, <strong>Switzerland</strong> stood out as one of the world’s strongest examples of accessibility in public transportation. Swiss Federal Railways (SBB) does more than provide audio announcements for visually impaired passengers. Through tactile guidance surfaces, accessible mobile applications, and Bluetooth beacon technology used in selected stations, passengers can identify their location more precisely, navigate to the correct platform, and complete transfers more independently.</p><p>I especially recommend exploring the <strong>SBB Inclusive</strong> application. It provides real-time station information, detects the station the user is in, confirms whether they have boarded the correct train, announces the next stop, and supports independent travel through VoiceOver. On selected trains, it can also use Bluetooth beacon technology to identify the carriage the passenger is in.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*XRsGODwq_kJYIKVY" /><figcaption>The app is designed to be easy to use, even with voiceover.(SBB Inclusive)</figcaption></figure><p>SBB Inclusive:<br><a href="https://www.sbb.ch/en/travel-information/apps/sbb-inclusive.html">https://www.sbb.ch/en/travel-information/apps/sbb-inclusive.html</a></p><p><strong>Sweden</strong>, on the other hand, takes a different but equally inspiring approach. The focus is not limited to public transportation; it extends to making the entire urban experience more accessible. GPS-based applications, audible traffic signals, accessible metro and bus announcements, tactile surfaces, and inclusive urban planning help visually impaired people move around cities more independently.</p><p>One particularly strong example is <strong>BlindSquare</strong>. As one of the most widely used accessible GPS applications, BlindSquare combines the user’s location with OpenStreetMap and Foursquare data to describe the surrounding environment through audio. It continuously provides information about nearby intersections, public transport stops, points of interest, and the direction of the destination, helping visually impaired people navigate cities more independently.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FJqU_jM_vkqc%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DJqU_jM_vkqc&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FJqU_jM_vkqc%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/88f99406ba0f30783e33168e7ae5be46/href">https://medium.com/media/88f99406ba0f30783e33168e7ae5be46/href</a></iframe><p>BlindSquare:<br><a href="https://www.blindsquare.com/">https://www.blindsquare.com/</a></p><p>Switzerland and Sweden may use different technologies, but their common goal is the same:</p><p>Helping people live without having to depend on others.</p><p>To me, that is one of the purest definitions of good user experience.</p><p>This research also reminded me of another important point. In UX, we do not want users to make unnecessary clicks. We simplify forms as much as possible, make information architecture easier to understand, and design user flows so they can be completed more quickly. We know that even small improvements can have a significant impact on the overall experience.</p><p>Shouldn’t we apply the same mindset to accessibility?</p><p>Many digital products still create readability problems because of insufficient color contrast. Some cannot be used with a keyboard, while others are not sufficiently optimized for screen readers, making it difficult for visually impaired users to use them independently. Images without alternative text and complex information architecture make these experiences even more difficult.</p><p>These are not simply technical shortcomings. They are design decisions that directly affect user experience.</p><p>And accessibility is not only an ethical responsibility. According to the <strong>World Health Organization</strong>, approximately <strong>1.3 billion people</strong>, around <strong>16% of the world’s population</strong>, live with some form of disability. This shows that accessibility is not an additional feature intended for a small group of users. It is a fundamental design principle that affects roughly one in six people worldwide.</p><p>Accessible design also benefits more than people with permanent disabilities. It creates better experiences for older adults, people with temporary injuries, people using their phones in low-light conditions, and users with many different needs and circumstances.</p><p>In other words, accessible design is better design for everyone.</p><p>When I left the event, what stayed with me was not a new technology or another regulation.</p><p>It was a visually impaired participant expressing a need that is as fundamental as anything the rest of us expect:</p><blockquote><strong><em>“I just want to be able to buy the product I want, like everyone else, without having to depend on someone.”</em></strong></blockquote><p>That was the moment I realized something else. As designers, we often think we are designing screens. We spend hours working on buttons, icons, cards, user flows, and conversion rates. But what we are actually designing is much more than an interface. We are shaping the way people interact with products, services, cities, and each other.</p><p>Good design is not only about aesthetics or usability. It is also about removing the invisible barriers people encounter in everyday life. Accessibility is not simply about adding alternative text to a website, improving contrast ratios, or ensuring screen reader compatibility. Those are important steps, but the real goal is to enable people to live their daily lives without having to depend on others.</p><p>When someone can choose the product they want in a supermarket, board a bus safely, complete a banking transaction without assistance, or use a digital service with the same level of ease as everyone else, that is when we can see the true impact of design. The success of design should not be measured only by the interfaces we see on a screen, but also by the independence it enables in people’s lives.</p><p>Perhaps the most important question we should ask ourselves is:</p><p><strong>Is the product, service, or experience we are designing truly accessible to everyone?</strong></p><p>If we cannot confidently answer “yes,” then there is still a great deal of work to do.</p><p>Because accessibility is not merely a feature of digital products. It is a foundation of human-centered design.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ea4e1f617a13" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/accessibility-is-not-a-feature-it-is-the-foundation-of-design-ea4e1f617a13">Accessibility Is Not a Feature, It Is the Foundation of Design</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Vibe Coding Isn’t About Not Coding. It’s About Building Faster.]]></title>
            <link>https://medium.com/space-ux-design/vibe-coding-isnt-about-not-coding-it-s-about-building-faster-42f3440fb141?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/42f3440fb141</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai-tools]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Sat, 01 Aug 2026 09:10:24 GMT</pubDate>
            <atom:updated>2026-11-06T13:28:47.168Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pGl9e30fVumey4ZtP11mBQ.png" /></figure><blockquote><em>A product designer’s perspective after shipping AlphaVest with AI.</em></blockquote><p>A few months ago, I never imagined I could design, develop, deploy, and document an enterprise FinTech application in just a few days instead of weeks.</p><p>Today, that’s exactly how I build products.</p><p>This new way of working is often called <strong>Vibe Coding</strong>.</p><p>Some people see it as hype, while others see it as the future.</p><p>After building <strong>AlphaVest</strong>, I realized the truth is somewhere in between.</p><p>Vibe Coding isn’t replacing software engineering. It’s changing how we collaborate with AI to build better products, faster.</p><h4>What is Vibe Coding?</h4><p>The term was introduced by <strong>Andrej Karpathy</strong> in 2025 to describe an AI-first way of building software. Instead of writing every line of code manually, you describe your intent, let AI generate an implementation, evaluate the result, refine it, and repeat. The developer becomes more of a product director than a code typist. <a href="https://www.ibm.com/think/topics/vibe-coding?utm_source=chatgpt.com">IBM</a></p><p>The important part is this:</p><p>You’re not asking AI to “build an app.”</p><p>You’re continuously guiding it.</p><p>Think of it as product design applied to software development.</p><h4>The Workflow Looks Surprisingly Similar to UX Design</h4><p>As designers, we’re already familiar with this way of working. We research, prototype, test, and iterate until we find the right solution.</p><p>Vibe Coding brings that same iterative mindset into development. Instead of creating another Figma screen, you’re building another working feature.</p><p>The feedback loop becomes much shorter, allowing you to validate ideas, learn faster, and improve the product continuously.</p><h4>How I Built AlphaVest</h4><p>AlphaVest started as a portfolio project.</p><p>The goal wasn’t just to design another dashboard.</p><p>I wanted to prove that a designer could take an idea all the way to production using modern AI tools.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rDCftn9qRT3vUq42jzH_7w.png" /><figcaption><strong>Vercel Live Production App: </strong>AlphaVest</figcaption></figure><p>My workflow looked something like this:</p><h4>1. Product Thinking First</h4><p>Everything started with product strategy. I used <strong>Google Stitch</strong> to explore ideas and quickly generate early product concepts before moving into detailed planning. From there, I defined the users, clarified the business goals, mapped the user flows, prioritized features, and planned the design system before writing a single line of code. AI didn’t replace this work. It simply helped me move through it faster.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e5KfYGwjlDT7wFtczBva7Q.png" /><figcaption>Google Stitch</figcaption></figure><h4>2. Figma Before Code</h4><p>Before writing prompts, I designed the entire experience in Figma. Every screen, interaction, and component became part of the product specification. Having a clear design made it much easier for AI to generate consistent, high-quality code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I6S1YE6u4gpkDNscn3DL1Q.png" /><figcaption>Figma and Design System</figcaption></figure><h4>3. Claude Became My Development Partner</h4><p>Instead of asking Claude to “build a dashboard,” I treated it like another engineer on the team. I shared the project architecture, design system, component hierarchy, accessibility requirements, responsive behavior, and edge cases. The better the context, the better the output. I quickly learned that good prompts create good software.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_SfkbTzZTbfxABxJkTlQpg.png" /><figcaption>Claude</figcaption></figure><h4>4. Small Iterations</h4><p>One of the biggest lessons I learned was to avoid asking AI to build everything at once. Instead, I built one feature, tested it, refined it, committed the changes, and then moved on to the next task. Working in small iterations made debugging much easier and felt very similar to how experienced product teams build software.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O_vYekVK63VSQRHDj6Wm_Q.png" /><figcaption>Google Antigravity</figcaption></figure><h4>5. Human Review Never Disappears</h4><p>One of the biggest misconceptions about Vibe Coding is that AI can do everything on its own. In reality, every feature still needs human review. I checked the UX consistency, accessibility, performance, responsiveness, component structure, and user flows before considering anything complete. AI generated the code, but I was still responsible for every decision.</p><h4>What Actually Changed?</h4><p>The biggest improvement wasn’t how fast I could write code, but how quickly I could make decisions.</p><p>Instead of spending hours building an idea just to see if it worked, I could validate it within minutes.</p><p>That made me more willing to experiment, test new ideas, and discard the ones that didn’t work.</p><p>The biggest surprise was that AI didn’t reduce iteration. It encouraged me to iterate even more.</p><h4>The Skills That Matter Now</h4><p>People often ask whether designers need to learn coding. I think the better question is which skills become more valuable in an AI-first world.</p><p>For me, those skills include product thinking, UX research, systems thinking, prompt engineering, information architecture, design systems, critical thinking, testing, and communication.</p><p>Most of these aren’t programming skills. They’re decision-making skills, and they become even more important when working with AI.</p><h3>Where Vibe Coding Fails</h3><p>Vibe Coding isn’t perfect, and AI still makes mistakes. It can over-engineer simple features, introduce unnecessary complexity, create inconsistent architectures, forget previous decisions, or confidently generate incorrect code.</p><p>Using AI-native development tools such as <strong>Google Antigravity IDE</strong> can improve the development experience by providing better context and AI-assisted workflows, but they don’t replace human judgment.</p><p>The quality of the output still depends on the quality of the instructions and feedback you provide. Human review remains essential to ensure the final product is maintainable, secure, and production-ready.</p><p><strong>Learn more about Google Antigravity IDE:</strong><br><a href="https://antigravity.google/product/antigravity-ide">https://antigravity.google/product/antigravity-ide</a></p><h4>My Biggest Lesson</h4><p>Building AlphaVest taught me that success isn’t about having smarter AI, it’s about learning how to communicate with it more effectively.</p><p>The future isn’t about replacing designers or developers.</p><p>It’s about combining product thinking, design, and AI into one seamless workflow.</p><p>To me, Vibe Coding means staying in control while giving AI clear direction to help bring ideas to life faster.</p><h4>Final Thoughts</h4><p>I don’t think Vibe Coding is just another trend because it’s changing how we build digital products.</p><p>AI doesn’t replace creativity. It gives us more time to focus on solving real problems.</p><p>Designers, developers, and founders can turn ideas into products faster than ever before.</p><p>The gap between an idea and a working product keeps getting smaller.</p><p>In the end, success isn’t about writing more code. It’s about building better products through better decisions.</p><p>That’s what Vibe Coding means to me.</p><p><strong>What are your thoughts on Vibe Coding?</strong> Have you started integrating AI into your design or development workflow, or are you still approaching it with caution? I’d love to hear your experience.</p><h4>Connect with Me</h4><p>💼 <strong>LinkedIn:</strong> <a href="https://www.linkedin.com/in/fatmadoganseckin/">https://www.linkedin.com/in/fatmadoganseckin/</a></p><p>🐙 <strong>GitHub:</strong> <a href="https://github.com/asparagam">https://github.com/asparagam</a></p><p>🎨 <strong>AlphaVest Case Study:</strong> <a href="https://www.figma.com/design/8pfyQnxH4X0iF7WzSsQ23F/AlphaVest">https://www.figma.com/design/8pfyQnxH4X0iF7WzSsQ23F/AlphaVest</a></p><p>🚀 <strong>Live Demo:</strong> <a href="https://alphavest-app.vercel.app/">https://alphavest-app.vercel.app/</a></p><p>If you enjoyed this article, <strong>follow me on Medium and connect with me on LinkedIn</strong>. I regularly share insights on AI Product Design, UX, Vibe Coding, and building AI-powered products in public.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=42f3440fb141" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/vibe-coding-isnt-about-not-coding-it-s-about-building-faster-42f3440fb141">Vibe Coding Isn’t About Not Coding. It’s About Building Faster.</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Designing Screens. Start Designing AI Systems.]]></title>
            <link>https://medium.com/space-ux-design/stop-designing-screens-start-designing-ai-systems-7afdb41be34f?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/7afdb41be34f</guid>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Thu, 09 Jul 2026 07:36:29 GMT</pubDate>
            <atom:updated>2026-11-06T13:29:28.713Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Why the next generation of Product Designers won’t be judged by pixels, but by the systems they build.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dHA1VJkA9i91aAgvOvorwg.png" /></figure><p>Most designers are still designing interfaces.</p><p>Buttons.<br> Cards.<br> Dashboards.<br> Navigation.</p><p>Meanwhile, AI-native companies are designing something entirely different.</p><p>They’re designing <strong>systems</strong>.</p><p>The biggest shift happening in product design isn’t about generating UI faster with AI. It’s about creating products that continuously improve through feedback, memory, evaluation, and user behavior.</p><p>The future of UX isn’t static interfaces.</p><p>It’s adaptive systems.</p><h3>The Three Eras of Product Design</h3><h4>Era 1 — Interface Design</h4><p>The goal was simple.</p><p>Make software usable.</p><p>Designers focused on layouts, typography, accessibility, and visual hierarchy.</p><p>Success meant users could complete tasks efficiently.</p><h4>Era 2 — Experience Design</h4><p>Products became ecosystems.</p><p>Design expanded beyond individual screens into complete customer journeys.</p><p>Research, personalization, onboarding, behavioral psychology, and retention became critical.</p><p>We stopped asking:</p><p><em>“Can users use this?”</em></p><p>We started asking:</p><p><em>“Will they come back?”</em></p><h4>Era 3 — System Design</h4><p>AI changes the question again.</p><p>Now we ask:</p><p><em>“Can this product learn from every interaction?”</em></p><p>The interface is no longer the product.</p><p>The system behind it is.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rLzNNjbCxtP-QusSV_HJig.png" /></figure><h3>The Four Layers of an AI Product</h3><p>Most successful AI products are built across four connected layers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tj_RfK4W_zD7t6d7ufxUdA.png" /></figure><h4>Layer 1 — Intelligence</h4><p>This is the model.</p><p>Claude.<br> GPT.<br> Gemini.<br> Open-source LLMs.</p><p>The model is only one component.</p><p>Choosing the most powerful model rarely creates the best product.</p><h4>Layer 2 — Orchestration</h4><p>The product needs structure.</p><p>Tasks must be routed.</p><p>Context must be managed.</p><p>Agents need specialized responsibilities.</p><p>Examples include:</p><ul><li>Planning</li><li>Research</li><li>Verification</li><li>Execution</li><li>Reflection</li></ul><p>Great AI products don’t rely on one giant prompt.</p><p>They coordinate many smaller decisions.</p><h4>Layer 3 — Memory</h4><p>Without memory, every conversation starts over.</p><p>Products become dramatically more useful when they remember:</p><ul><li>User preferences</li><li>Previous decisions</li><li>Successful workflows</li><li>Failed attempts</li><li>Project history</li></ul><p>Memory creates continuity.</p><p>Continuity creates trust.</p><h4>Layer 4 — Evaluation</h4><p>This is the layer most teams skip.</p><p>AI should never assume it’s correct.</p><p>Instead, products should constantly ask:</p><ul><li>Did the output solve the user’s problem?</li><li>Is the response complete?</li><li>Does it match the design principles?</li><li>Can it be improved?</li></ul><p>Products that evaluate themselves improve faster than products that simply generate answers.</p><h3>AI Products Are Feedback Loops</h3><p>Traditional software follows this pattern:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M3ZRhBSBbeO8RZb5If7uZg.png" /></figure><p>Every interaction becomes training data for the product experience — even if the underlying model never changes.</p><p>That’s where competitive advantage comes from.</p><h3>Why UX Designers Matter More Than Ever</h3><p>Many people assume AI reduces the importance of UX.</p><p>The opposite is happening.</p><p>As models become increasingly capable, differentiation shifts away from intelligence and toward experience.</p><p>Users don’t remember which model powered the product.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_EHMUlHfsEeH-nk6mbVMIQ.png" /></figure><p>They remember:</p><ul><li>whether it understood them</li><li>whether it saved time</li><li>whether it made good decisions</li><li>whether it felt trustworthy</li><li>whether it improved over time</li></ul><p>Those are UX problems.</p><p>Not AI problems.</p><h3>Designing AI Means Designing Behavior</h3><p>Think about products like Spotify.</p><p>The recommendation model matters.</p><p>But the experience users remember is discovering the right song at exactly the right moment.</p><h3>A Practical Framework for AI Product Teams</h3><p>When designing an AI-powered feature, ask these six questions:</p><ol><li>What should the AI decide?</li><li>What should the user always control?</li><li>What should the product remember?</li><li>How will the system verify its own output?</li><li>What feedback improves future interactions?</li><li>What metrics prove the experience is actually getting better?</li></ol><p>If your product can’t answer these questions, you’re probably building a chatbot — not an AI system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6kCfhF-0dNO1CqhKhtZfEA.png" /></figure><h3>The Biggest Mistakes Teams Make</h3><ul><li>Shipping a powerful model with no memory</li><li>Treating prompts as product architecture</li><li>Measuring response quality instead of user outcomes</li><li>Ignoring evaluation and feedback loops</li><li>Designing conversations instead of workflows</li><li>Optimizing for demos instead of long-term behavior</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DLc7qpxM89bRV-Lv0Y2Iag.png" /></figure><h3>The Future of Product Design</h3><p>The best AI products won’t win because they use the smartest model.</p><p>They’ll win because they create the smartest systems.</p><p>The role of designers is changing from crafting interfaces to shaping intelligence.</p><p>From designing screens to designing decision-making.</p><p>From building features to building systems that continuously improve.</p><p>That’s the next evolution of UX.</p><p>And it’s already here.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7afdb41be34f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/stop-designing-screens-start-designing-ai-systems-7afdb41be34f">Stop Designing Screens. Start Designing AI Systems.</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why the future of design belongs to strategists, systems thinkers, and AI-native product builders]]></title>
            <link>https://medium.com/space-ux-design/why-the-future-of-design-belongs-to-strategists-systems-thinkers-and-ai-native-product-builders-dda893cd5558?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/dda893cd5558</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[design-strategy]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[growth-design]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Tue, 09 Jun 2026 06:34:58 GMT</pubDate>
            <atom:updated>2026-11-06T13:28:58.207Z</atom:updated>
            <content:encoded><![CDATA[<p>The latest <em>State of AI in Design</em> report confirms a major shift: AI is no longer just a productivity tool. It is becoming part of the product development operating system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ison-S5TvkzkLezsB5y1-g.png" /><figcaption>Al-generated cover image illustrating the evolution of product design toward Al-native workflows, growth strategy, and scalable systems.</figcaption></figure><p>For years, design focused on interfaces, user flows, and visual systems. Today, AI can generate many of those outputs in seconds. As execution becomes easier, strategy becomes more valuable.</p><p>The question is no longer:</p><p><em>“Which AI tool should I use?”</em></p><p>It’s:</p><p><em>“How do I build workflows where AI becomes part of the entire product lifecycle?”</em></p><h3>From Tools to Systems</h3><p>Designers now work with AI copilots, coding assistants, research tools, automation platforms, and rapid prototyping systems.</p><p>The real advantage does not come from using these tools individually. It comes from connecting them into scalable workflows.</p><p>Leading companies are not just adding AI features. They are redesigning how products are built.</p><p>Companies like Perplexity, Anthropic, Notion, Linear, and Duolingo share a common pattern: they use AI to improve outcomes, not simply increase productivity.</p><p>The shift is clear. AI is moving from a collection of tools to an integrated part of how products are designed, built, tested, and optimized.</p><h3>Why Growth Design Matters More Than Ever</h3><p>As AI reduces the effort required to create screens, product teams are focusing on what drives business value:</p><ul><li>Activation</li><li>Retention</li><li>Conversion</li><li>Engagement</li><li>Revenue</li></ul><p>This is why Growth Design and UX Strategy are becoming critical skills.</p><p>Future designers will be measured less by outputs and more by outcomes. The ability to understand user behavior, identify friction points, and influence business metrics is becoming a key differentiator.</p><p>Creating interfaces is becoming easier. Creating impact is not.</p><h3>Designing Adaptive Experiences</h3><p>Traditional UX relied on static journeys. Every user moved through the same onboarding flow, navigation structure, and experience.</p><p>AI changes that model.</p><p>Products can now adapt in real time based on user intent, behavior, context, and preferences.</p><p>Designers are no longer creating a single user flow. They are designing systems that can generate personalized experiences at scale.</p><p>The challenge is shifting from interface design to decision design.</p><p>The products that win in the AI era will not be the ones with the most features. They will be the ones that best adapt to individual user needs.</p><h3>The Rise of the Strategic Product Designer</h3><p>As AI takes on more execution work, the role of the designer expands.</p><p>The next generation of designers will combine:</p><ul><li>Product Thinking</li><li>UX Strategy</li><li>Growth Optimization</li><li>Systems Design</li><li>Data-Informed Decision Making</li><li>AI-Native Workflows</li></ul><p>These skills sit closer to business strategy than traditional design execution.</p><p>The most valuable designers will not be those who can create screens the fastest. They will be those who can connect user needs, business goals, and emerging technologies into scalable product systems.</p><h3>Final Thought</h3><p>AI is not replacing designers. It is raising the value of what designers do best.</p><p>As execution becomes automated, skills like strategy, systems thinking, product judgment, creativity, and understanding human behavior become even more important.</p><p>The future is bright for designers who embrace AI and evolve with it.</p><p>This is not the end of design. It is the beginning of a bigger role.</p><h3>References</h3><p>This article is inspired by insights from the State of AI Design 2026 report, based on research with hundreds of designers and design leaders exploring how AI is transforming the future of product development.</p><p>Read the full report:</p><p><a href="https://stateofaidesign.com/">https://stateofaidesign.com/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dda893cd5558" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/why-the-future-of-design-belongs-to-strategists-systems-thinkers-and-ai-native-product-builders-dda893cd5558">Why the future of design belongs to strategists, systems thinkers, and AI-native product builders</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Pragmatic UX Research: Choosing the Right Methods When Time and Resources Are Limited]]></title>
            <link>https://medium.com/space-ux-design/pragmatic-ux-research-choosing-the-right-methods-when-time-and-resources-are-limited-2f3a54440d38?source=rss-f427704d22c0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f3a54440d38</guid>
            <category><![CDATA[user-interviews]]></category>
            <category><![CDATA[ux-research]]></category>
            <category><![CDATA[research-methods]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[usability-testing]]></category>
            <dc:creator><![CDATA[Fatma Dogan Seckin]]></dc:creator>
            <pubDate>Sun, 07 Dec 2025 11:35:10 GMT</pubDate>
            <atom:updated>2026-11-06T13:28:24.920Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>In fast moving product teams, designers and researchers are expected to deliver strong insights with very limited time, budget, or research capacity. This challenge is clearly described in Wolfram Nagel’s article titled Pragmatic UX Research: Doing More With Less, and it reflects a common reality in UX teams today.</blockquote><p>This article introduces practical UX research methodologies, how to pick the right ones at the right moment, and how a pragmatic mindset helps you create meaningful impact even with tight constraints.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lDx9mz9eEAtPaP-aSo3o8w.png" /></figure><h4>Why Pragmatic UX Research Matters</h4><p>Teams often face one or more of the following problems:</p><ul><li>No dedicated UX researcher</li><li>Tight deadlines and ambitious roadmaps</li><li>Limited access to users</li><li>Stakeholders expecting quick data driven decisions</li></ul><p>A pragmatic approach focuses on doing fewer things, but doing them with higher precision. The goal is not to build a perfect academic research system but to create a sustainable and repeatable process that supports product decisions.</p><h4>UX Research Methodologies: A Practical Guide</h4><p>Understanding motivations, behaviors, and context</p><p>Qualitative methods answer the question Why. They help you uncover deep insights and real user needs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lEIGK4eP049-DfgCgdJmOw.jpeg" /><figcaption><a href="https://unsplash.com/photos/person-leaning-on-table-QEsaXprgVfQ"><strong>Unsplash</strong></a></figcaption></figure><h4>1. Qualitative Research</h4><blockquote><strong>Understanding motivations, behaviors, and context:</strong> Qualitative methods answer the question Why. They help you uncover deep insights and real user needs.</blockquote><h4>User Interviews</h4><p>Ideal for exploring motivations, frustrations, emotional triggers, and decision making patterns.</p><h4>Contextual Inquiry or Ethnographic Research</h4><p>Observation of users in their natural environment. Useful to understand what users actually do rather than what they say.</p><h4>Diary Studies or Longitudinal Studies</h4><p>Good for understanding long term behaviors, routines, and evolving experiences.</p><h4>Card Sorting</h4><p>Helps designers understand how users group information and how they expect navigation to work.</p><p><strong>When to use: </strong>Discovery stage or early concept definition.</p><h4>2. Quantitative Research</h4><blockquote><strong>Measuring what users do:</strong> Quantitative methods help you validate assumptions and measure behavior at scale. They answer questions like <strong>How many, How often, and How much.</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0wYrd6ZVGkYT0uMOKr78cQ.jpeg" /><figcaption><a href="https://unsplash.com/photos/person-using-macbook-pro-pypeCEaJeZY"><strong>Unsplash</strong></a></figcaption></figure><h4>Surveys</h4><p>Structured questions sent to large user groups to collect measurable insights.</p><h4>Analytics and Behavioral Data</h4><p>Tools such as GA4, Amplitude, Mixpanel or Yandex Metrica reveal user actions such as clicks, flows, drop offs, and time spent.</p><h4>A or B Testing</h4><p>Compares different design variations based on measurable results such as conversion rate or engagement.</p><p><strong>When to use:</strong> After launch or during optimization cycles.</p><h4>3. Mixed and Evaluative Methods</h4><p>Combining depth and scale</p><p>These methods bring qualitative and quantitative insights together. They are highly effective in real product development cycles.</p><h4>Usability Testing</h4><p>Shows how easily users can complete tasks. Reveals friction points, confusion, and usability failures.</p><h4>Heuristic or Expert Review</h4><p>UX experts identify issues using established usability principles. This method is fast and ideal when access to users is limited.</p><h4>Concept Testing</h4><p>Used to validate early ideas, flows, sketches or wireframes before deeper investment.</p><p><strong>When to use: </strong>During prototyping, pre launch phases, and continuous improvement cycles.</p><h4>Choosing the Right Method for Each Stage</h4><h4><strong>Discovery Stage:</strong></h4><ul><li>User interviews</li><li>Contextual inquiry</li><li>Diary studies</li><li>Open card sorting</li></ul><h4>Define and Structure Stage:</h4><ul><li>Card sorting</li><li>Tree testing</li><li>Field observation</li></ul><h4>Design and Prototype Stage</h4><ul><li>Concept testing</li><li>Usability testing</li><li>Heuristic review</li></ul><h4>Launch and Optimize Stage</h4><ul><li>Analytics</li><li>Surveys</li><li>A or B testing</li><li>Funnel analysis</li></ul><p>This simple framework helps teams stay aligned and ensures research is applied continuously instead of occasionally.</p><h4>The Pragmatic Research Mindset</h4><p>Wolfram Nagel highlights three core principles of pragmatic UX research.</p><h4>1. Prioritize impact over method variety</h4><p>Choose methods that answer your question directly. Do not use every method. Use the right method.</p><h4>2. Combine fast expert methods with targeted user studies</h4><p>Start with a heuristic review</p><p>Validate with quick usability tests</p><p>Confirm with analytics</p><p>This saves time while keeping decisions grounded in user reality.</p><h4>3. Collaborate early and often</h4><ul><li>Share early drafts</li><li>Align with stakeholders</li><li>Keep your process lightweight, actionable, and easy to repeat</li></ul><h4>Additional Resources to Deepen Your UX Research Skills:</h4><ul><li><a href="https://maze.co/guides/ux-research/methods/">Maze UX Research Methods Guide</a></li><li><a href="https://nulab.com/learn/design-and-ux/ux-research-methods/">Nulab UX Research Methods Overview</a></li><li><a href="https://www.looppanel.com/blog/qualitative-ux-research-methods">LoopPanel Qualitative UX Research Articles</a></li><li><a href="https://www.uxtweak.com/ux-research/quantitative-vs-qualitative/">UXTweak Quantitative vs Qualitative Research</a></li></ul><h4>Conclusion</h4><p>Effective UX research does not require a large budget or a large team. It requires clarity, focus, and choosing methods based on your real constraints. By combining qualitative depth, quantitative validation, and practical evaluation techniques, product teams can make smarter and faster decisions while staying connected to user needs.</p><p>A pragmatic UX research approach helps you scale insights without losing quality, even in high pressure environments.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f3a54440d38" width="1" height="1" alt=""><hr><p><a href="https://medium.com/space-ux-design/pragmatic-ux-research-choosing-the-right-methods-when-time-and-resources-are-limited-2f3a54440d38">Pragmatic UX Research: Choosing the Right Methods When Time and Resources Are Limited</a> was originally published in <a href="https://medium.com/space-ux-design">Space UX Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>