Katie Schneider

Stop Trusting Fluent Answers. Start Designing Better Prompts. 🤔🚦💡 A practical guide to spot AI blind spots. And turn fluffy prompts into sharp, reliable asks. It’s called “Deconstructing an AI Conversation: A Pattern Recognition Guide.” It shows where LLMs overreach, and how to counter it. It helped refine prompts for clarity, context, and outcomes. 👉 What it is A three-part walkthrough that demystifies confident-but-wrong answers. It upgrades prompting from guesswork to craft. 👇 How it works ☝️ Section 1: The prompt and the flashy answer See how fluency masks missing context and shaky assumptions. ✌️ Section 2: The blind spots, clearly labeled False authority and reality gaps. Information assumptions and decision shortcuts. Position bias from attention mechanics. 🤟 Section 3: The fixes and prompt moves Ask for calibrated confidence. Force context gathering first. Make assumptions visible. Request trade-offs, not just answers. Why read it? Because better prompts reduce rework and raise decision quality. Designers and researchers will get immediate mileage. Thanks to Tey Bannerman for the excellent work and sharing this guide. It’s worth it because it turns patterns into practice in minutes. Follow me to stay updated on the Empathic Web, AI, and design. #UXDesign #UX #Design #PromptEngineering #LLM #AIPractices #CognitiveBias #ProductDesign #EthicalAI

4

The industry changed. How we work, what companies expect, even what "being ready" means. But here's the thing: You don't need another course or bootcamp. You need practice with real designers who get it. That's why we built Practice Round. Free mock interviews, case study workshops, and honest feedback from designers who've been where you are. Join the 70+ designers already practicing together → practiceround.co

3

We’re entering a period of structural change in how product design and development teams operate. As AI accelerates production and organizations shift toward outcomes, the role of UX is expanding beyond execution into strategy, orchestration, and decision-making. I’ve been exploring how those dynamics are evolving: from outputs to outcomes, silos to journeys, and production to orchestration. This is part 1 of a 3-part series on how UX, product, and engineering teams are changing in the agentic era. https://lnkd.in/eQndDQMe #designexcellence #agenticAI #UXleadership #designleadership

39

What if the next big design breakthrough does not come from a senior leader, but from a new grad who just mastered the latest AI tool? There’s a growing gap in design hiring. Most openings are for very senior roles, while younger designers just graduating are struggling to break in. But here’s the reality: younger designers bring something we desperately need. They are fluent in AI, eager to adopt new processes, and constantly tuned into the tools and methods shaping design today. They are open to learning, open to change, and quick to experiment. If we dismiss them, we are missing out on fresh perspectives and approaches that can elevate our practice. Established methods are not automatically the best methods. The best design teams are not built only on experience. They thrive on diversity of perspective. Senior and junior designers together create balance: depth meets agility. I recently hired a young contractor who dove headfirst into the AI tools we were piloting at our company. He introduced me to platforms like Lovable and UX Pilot, among others tools that I later shared with our senior design leadership. Within just three months, his curiosity and initiative created ripple effects across our org. He didn’t just adapt to our systems; he helped evolve them. To recruiters and hiring managers, give younger designers the chance to show their work. You may find they are not just ready for the industry, they are already shaping its future. And for those younger designers trying to break in, I know it is tough right now. If you need a second set of eyes on your portfolio, a bit of career advice, or even just encouragement, my inbox is open. Sometimes all it takes is one conversation to shift your path forward. #UXDesign #DesignCareers #DesignLeadership #AIinDesign #Mentorship

25

1 Comment

AI-Native Design Systems: From Component Libraries to Machine-Readable Architecture —————- Your design system is no longer documentation. It's an instruction set. For a decade, design systems existed to align humans — tokens, components, patterns that designers and developers could reference and interpret. AI changed the contract. When Claude Code, Cursor, or v0 consumes your design system, it doesn't read it the way a designer does. It executes it. And execution is unforgiving in ways that human interpretation never was. A human designer understands that "modern and clean" means restraint, whitespace, probably a sans-serif. They resolve ambiguity through judgment. An AI agent resolves ambiguity through generation. That distinction is everything. What makes a design system AI-native? Three properties separate systems that govern AI behavior from systems that AI ignores: 1. Structure over description color/primary/600, radius/md, spacing/3 spacing/4 is a specification. "The button should feel confident and accessible" is not. One can be executed. One cannot. 2. Boundaries over inventory Traditional systems document what exists. AI-native systems define what is permitted. Without explicit boundaries, AI invents. With them, AI composes. 3. Translation fidelity If your system lives cleanly in Figma but fractures in JSON or CSS variables, you don't have an AI-native system. You have a human-readable artifact that breaks under machine conditions. The failure most teams don't see coming: Encoding inconsistency. If your live product has seven button variants — three of which are historical accidents — and you formalize all seven, AI will generate all seven with equal confidence. You haven't built a system. You've institutionalized the mess. Normalize before you formalize. Always. The deeper shift: In AI-native workflows, designers are no longer producing screens. They're engineering the rules that generate screens. That's a different cognitive skill. Closer to language design than visual composition. The deliverable isn't the artifact — it's the constraint set that makes coherent artifacts possible at scale. The teams that understand this will use AI as a multiplier. The rest will keep writing better prompts and wondering why consistency never holds. Design systems that can't instruct a machine precisely won't survive the next phase of product development. If you’d like to read the full article — including the architectural model, failure taxonomy, and governance implications — you can find it here: https://lnkd.in/dK7E9dtK #AINativeUX #DesignSystems #AIDesign #ProductDesign #SystemDesign #UXLeadership #GenerativeAI

8

Thank you to everyone who joined UXPA Seattle today for our UX strategy session with Jaime Levy. Jaime shared a thoughtful, real-world case study illustrating how UX strategy and discovery practices can be applied to everyday problem-solving. It was a great reminder that UX strategy isn’t just for products and teams: it’s a way of thinking that applies well beyond work contexts.  #UXStrategy #UXCommunity #UXPA #ProductThinking

3

A lot of this reflects my own personal view of how the design process is changing, and its effects on the archetypes of designers needed for the next chapter

19

This hit close to home. This quarter we've been actively rethinking our design process. Removing rituals, pulling back on some discovery activities, and prioritizing speed and learning over craft and critique. It can be uncomfortable but it's the right direction. This episode puts a lot of it into words.

27

7 Comments

Big news: Anthropic released their competitor to Google Stitch and Figma Make, aptly named Claude Design. I spent a bit of time today testing it out, and my initial reaction is that this is a next evolution of vibe coding tools and it takes them to the next level. On the surface, it looks very similar: prompt to build, refine, and evolve what the system interprets from your provided documentation. The piece that takes this to the next level is the generative UI capability they call "knobs:" which is a dynamic palette of controls that changes based on what you're editing. Click on a piece of text, and it shows you all the ways you can change that text element's styling. Click on a dynamic map, and it lets you change the basemap styles, map markers, legend, or whatever else. This is the key innovation: a real, practical use case for generated interfaces. You no longer need to fight to edit with prompts like in Figma Make, v0, or Lovable wasting tokens, or with limited/generic UI editors like in Cursor. Claude Design lets you fine tune anything you prompt to life, and that level of control is a game-changer. https://lnkd.in/edsPQYmV

58

4 Comments

Product designers aren't just making things pretty. They're secret weapons for rapid product validation. 🚀 Last month, I talked to a PM and he told me that he spent 6 weeks building features based on assumptions. The result? Users hated it. Complete rebuild required. ⏰💸 But here's what he figured out: UX/UI isn't decoration. It's validation currency. 💰 Here's how design accelerates your product decisions: 🎯 RAPID PROTOTYPING = INSTANT FEEDBACK Skip months of development. Build interactive prototypes in days. Test with real users before writing a single line of code. 📊 USER JOURNEY MAPPING = ASSUMPTION KILLER Designers force you to walk in your users' shoes. Every pain point becomes visible. Every assumption gets challenged. 🔄 A/B TESTING MADE VISUAL Two design variations = two hypotheses tested simultaneously. Data-driven decisions in weeks, not quarters. ⚡ STAKEHOLDER ALIGNMENT AT LIGHT SPEED Nothing sells an idea like seeing it. Mockups get buy-in faster than any PowerPoint deck ever will. The math is simple: 2 weeks of design exploration > 2 months of wrong product development 📈 I've seen PMs cut validation cycles from 3 months to 3 weeks just by partnering with design from day one. Not after requirements. Not after wireframes. From the first "what if" conversation. 🤝 Your users don't care about your backlog priorities. They care about whether your product actually solves their problem. Design helps you find out before you build. 🎯 What's your biggest challenge in validating product ideas quickly? Drop it in the comments - let's solve it together! 💬 -------------------- 🙋🏻‍♂️ Hi there! I'm Matías Sandoval ‍🧑🏻‍💻 I'm a Product | UX/UI designer 🤩 Helping brands craft experiences that users love 🔔 Follow me for more content ♻ If you found this helpful, please share this post 😉 💼 Let's work together on Contra: https://lnkd.in/e-5tXMUx

4

While I don't appreciate the typical clickbait stating that "Design is Dead," I found this video to be really insightful. The classic design process is too rigid and outdated for the modern ecosystem of tools. While I still think the old process still offers a lot of context and wisdom, it's time to think differently. #DontrusttheProcess https://lnkd.in/gjsfuY9X

1

I’d like to connect with other design leaders shaping design for LLMs and AI-generated interfaces. If you’re architecting how design thinking translates into machine-readable design logic and making design AI-operable, feel free to send me a direct message. I’d love to discuss insights and approaches 🤝

9

3 Comments

Eight years of using Figma, and I'm still missing this one small thing Sketch has. When you resize a layer with math (e.g. 128*2), Sketch lets you add a direction modifier behind it (L/R/T/B/C/M) to tell it which way the layer should grow. So 128*2B resizes from the bottom. Your element stays in place. No extra clicks to reposition it. Figma still doesn't have this. Is it a big deal? Probably not on its own. But when you're deep in polishing, nudging specs, tightening layouts, these small frictions add up. And polishing is kind of where Figma still earns its spot in my toolbox. Not for everything, but for that final 20% of getting things right it's still the fastest tool I have. Which is exactly why gaps like this one sting. There's a lot of pressure on Figma right now to move fast on AI. Closing small gaps like this won't change that big picture but in the meantime, it could at least stay frictionless at the thing it's currently good at.

50

6 Comments

I've been doing a lot of design work lately in Claude. As someone who knows just enough to be dangerous about what "good" looks like, I don't think the role of the designer is going away, but like many roles, it will change. Here's what I think the designer's role becomes: someone who: - Brings vision, creativity, and taste to the work. - Rapidly prototypes with AI tools instead of building every version by hand. - Takes what AI produces and layers in the original point of view that makes a design actually unique. I feel like we can't all see it now, but very shortly, we will all look at pure AI-generated web design and cringe. We need human nuance to make it impactful.

23

8 Comments

Are IC enterprise designers unintentionally sabotaging their portfolios in this AI-obsessed market? If you’re a Product Designer updating your case studies… you’ll probably want to read this... I’m actively hiring two Sr-to-Lead Product Designers, and after reviewing a lot of portfolios recently, a pattern keeps repeating: 👉 Too many portfolios are leading with AI chatbot projects. Not sprinkled in… leading with them. Often multiple ones in the same portfolio as Case Study #1 and #2. Here’s the problem👇 AI chatbots are important, but they’re not the center of gravity for most enterprise / SaaS product teams. In the last six-nine months, many of these patterns are already shipped and commoditized. And now with agentic flows, the UI work becomes more about reviewing, validating, and confirming work AI has already completed behind the scenes. Which means… “Old school” multi-step, end-to-end product flows are still valid – with a twist. Because every IC job post right now is calling for “high craft” or “consumer-grade design” – it’s no longer acceptable for flows to simply “get the job done.” Hiring panels want to see refinement and polish in the final output.  Otherwise you’ll get the feedback we give far too often: “The design thinking is great… but the visual design is just… [sigh/meh]… not quite there.” Visual craft always matters. Everywhere. At every company. Whether D2C or B2B or Enterprise. So if you’re refreshing your portfolio, here’s my ask as someone hiring in this tough market: Don��t over-index on AI chatbots just because they’re hot right now. Show me your complex, nuanced, multi-stage flows—the kind that prove you can handle the real product work most teams need. Just make sure they’re polished (ie. That “high craft” buzzword again), intentional, and tight. Because that’s what will actually differentiate you in 2025 and into 2026. And now the proverbial 🔌: If you’re a high-performing Sr. or Lead looking to innovate in the HR and Comp tech space, holler at me in my DMs or Comments. Thanks! 🙏

36

3 Comments

This is just another symptom of big business going further down the path of short-term thinking; the note about "not this quarter's problem" is exactly right. Visionary leadership hasn't been in evidence at most of these companies for a long time. The hope of course is that at some point all these product are just so boring and interchangeable that a dark horse sweeps right past them, shaking up the field, kicking everyone's behind, and suddenly deep learning is interesting again. It might take awhile. There's enough profit going around to keep the big shots complacent for a bit. But make no mistake - touting a 996 schedule while doing the same thing as everyone else is complacency. And it's a short term point of view.

7

Why does it feel like designers are more focused on mastering technical workflows and endless AI tools than actually designing?

16

7 Comments

Claude just released Claude Design. I’ve spent a bit of time with it and it’s interesting so far. I like where it's headed. Over the past couple of years, I’ve worked pretty heavily with AI tools. Claude, OpenAI, Cursor, Lovable. I’ve even built product and custom skills to support parts of the design process. Some were helpful. Others felt like extra work without enough return. What I’m starting to like here is the shift toward a more integrated experience. Being able to connect a design system, make tweaks directly, and QA outputs in the same environment feels like a better balance. Less relying purely on prompts, more control over the outcome. That’s been the pattern for a while. Trying to fit tools into the process instead of the other way around. That's not to say that workflows can't change. This feels like a step in a better direction, but it’s still early. What stands out isn’t just what it can do, but how it fits. It’s starting to align more naturally with how we already think, design, and iterate. Less friction, more flow. AI can absolutely push us toward better directions faster and help carry craft further into the final output. But it doesn’t replace thinking. Without that human judgment and intent, it’s easy to land on average solutions that miss the mark. The tools are getting better. That’s clear. But the work is still ours. Craft, taste, and decision-making are what shape the outcome. And the outcome is what matters. Excited to see how this evolves and continues to change my workflow: https://lnkd.in/eqGvEBC3

12

1 Comment

“Comparison is the thief of joy.” As an IC designer I compared myself to other ICs. As a design manager I compared myself to other managers. None of this was helpful, and at times led me down the wrong path. No matter where you are in your career, there will always be someone better than you. Don’t let that influence your career decisions. Just keep developing your craft. I talk about these and more tips I learned from management in my latest newsletter. Link in comments below.

23

7 Comments