AI in Product Design

Product managers & designers working with AI face a unique challenge: designing a delightful product experience that cannot fully be predicted. Traditionally, product development followed a linear path. A PM defines the problem, a designer draws the solution, and the software teams code the product. The outcome was largely predictable, and the user experience was consistent. However, with AI, the rules have changed. Non-deterministic ML models introduce uncertainty & chaotic behavior. The same question asked four times produces different outputs. Asking the same question in different ways - even just an extra space in the question - elicits different results. How does one design a product experience in the fog of AI? The answer lies in embracing the unpredictable nature of AI and adapting your design approach. Here are a few strategies to consider: 1. Fast feedback loops : Great machine learning products elicit user feedback passively. Just click on the first result of a Google search and come back to the second one. That’s a great signal for Google to know that the first result is not optimal - without tying a word. 2. Evaluation : before products launch, it’s critical to run the machine learning systems through a battery of tests to understand in the most likely use cases, how the LLM will respond. 3. Over-measurement : It’s unclear what will matter in product experiences today, so measuring as much as possible in the user experience, whether it’s session times, conversation topic analysis, sentiment scores, or other numbers. 4. Couple with deterministic systems : Some startups are using large language models to suggest ideas that are evaluated with deterministic or classic machine learning systems. This design pattern can quash some of the chaotic and non-deterministic nature of LLMs. 5. Smaller models : smaller models that are tuned or optimized for use cases will produce narrower output, controlling the experience. The goal is not to eliminate unpredictability altogether but to design a product that can adapt and learn alongside its users. Just as much as the technology has changed products, our design processes must evolve as well.

Design for AI-native products changes the role of designers from building blueprints to shaping ecosystems. Traditional product design is like drafting a blueprint: predictable, linear, and structured. Designing for AI products, however, feels more like cultivating an ecosystem. It’s unpredictable and dynamic, requiring designers to embrace ambiguity. So how is designing for AI-native products different? 1. Designing for probabilities, not certainties: Traditional design assumes predictable outcomes. With AI, outputs vary based on data and context, so designers must create patterns for feedback and error handling that feels intuitive. 2. Design systems, not flows: AI products adapt over time, requiring modular systems that can handle continuous changes and scale. 3. Designing feedback loops: Users collaborate with AI to refine outcomes, making iteration cycles intuitive and efficient. Personalization features, like custom rules or GPT configurations, adds depth. 4. Evaluation criteria: AI needs evaluation frameworks based on to measure and improve accuracy and relevancy over time. This should be grounded in user needs and goals. 5. Considering the cost: Running AI has real costs, so designers must understand and optimize to balance user needs with business constraints. But some things remain the same. → User-centricity is timeless: Understanding user needs and pain points is still foundational. → Non-AI foundations matter: Onboarding, settings, IA, etc. remain critical for good product design. → Design systems are still your best friend: A strong design system saves time and ensures consistency, especially with AI’s unpredictability. Designing for AI-native products redefines what’s possible by combining innovation with empathy. I’m thrilled for the experimental patterns that will shape the future of design.

AI won’t just enhance existing products. It will rewrite the product playbook. That was the core thesis from Sarah Guo, founder of Conviction, in our recent #MicrosoftPMCon. Here are 5 takeaways that left a lasting impression—and how I’m thinking about them as a product maker: 1️⃣ Don’t just add AI. Reimagine the product. Sarah distinguishes between AI-enhanced and AI-native. The latter doesn’t bolt AI onto workflows—it starts with the assumption that the product is the intelligence. I’m learning to ask: If this were built today from scratch—with an LLM as a teammate—what would it look like? 2️⃣ The most valuable products won’t be wrappers—they’ll be rethinkers. We’re past the "GPT inside" phase. The next wave requires scaffolding, orchestration, user trust, and graceful failure handling. Sarah compares it to how Salesforce rewrapped relational databases. The new "CRM wrappers" will be intelligent, fluid, and role-specific—not bound by legacy UX. 3️⃣ User understanding matters even more in AI. It’s not just about prompt tuning. Sarah emphasized that deep user empathy is essential when building agents that make decisions, not just surface info. Her portfolio companies hire lawyers to co-design legal AI tools and clinicians to build in healthcare. That level of domain fluency is what differentiates useful from magical. 4️⃣ New muscle groups are required. AI product development means navigating long-run tasks, unpredictable outcomes, and non-deterministic behavior. The best teams are blending product, infra, and research to ask: How do we make a 5-hour task feel delightful? How do we teach users to debug their AI teammates? 5️⃣ Product builders now need a ‘research ear’. In most tech eras, you could ignore what was happening in the labs. Now? The frontier shifts every two months. Sarah put it bluntly: If you’re not projecting capability curves, you’re building for a world that no longer exists. Curious: What skill do you think today’s PMs need that didn’t matter 3 years ago? Let’s learn from each other. —  👋 Hi! I’m Shyvee, I share insights on AI and the future of work. Subscribe for AI insights, programs, and an invitation to our AI Enthusiast Community: https://lnkd.in/eR2ebrEM #ProductManagement #AI #MicrosoftLife

The AI shift is already happening: Product design isn’t just AI-assisted. It’s becoming AI-influenced. Tools like Figma Make, Galileo, Uizard, and v0.dev are no longer just helping us build—they’re starting to shape what gets built. The way we explore ideas, validate decisions, and even define MVPs is evolving fast. Designers aren’t just crafting UI anymore—we’re curating prompts, training workflows, and guiding machine-generated options toward better outcomes. This doesn’t replace creativity. It amplifies it. If we lean in with intention. The next era of product design will belong to teams who can blend systems thinking, UX intuition, and AI literacy. Not just use the tools—but design with them. Are you seeing this shift in your work?

Why I think the AI era is the best time for product designers?👉 The launch of Gemini 3 felt like a fast-forward button pressed on the design industry. For the first time, it’s clear: AI is no longer just a tool — it’s taking over the most basic, repetitive parts of design. This triggered a wave of anxiety among designers: “Will we be replaced?” “AI can even sketch — what’s left for us?” But after months of observing and experimenting, I’m seeing a different pattern emerge: 1️⃣The design workflow is being rewritten: The old linear process (research → sketch → wireframe → hi-fi → delivery) no longer holds. Designers are shifting from “how to make it” to “what to make, why it matters, and how far to push it.” 2️⃣Designers are evolving from executors to decision-makers As execution becomes cheap, two skills become priceless: (1) Problem framing: AI can draw, but it can’t tell you what’s worth drawing. (2) Taste & judgment: AI can generate 50 versions — choosing the right one requires insight, not execution. This shift is pushing designers closer to full-stack product strategy. Instead of anxiety, I believe this actually marks the start of 🚀the next golden 3 years for product people🚀 Opportunities I see: 👉 AI-Native Designers 👉 Deep design × product integration 👉 Multimodal experiences & new interaction paradigms 👉 Becoming the AI driver in your team (start small: workflows, guidelines, pilot projects) Last summer, I built a few vibe-coding prototypes and ended up running a month of AI training for the team — small experiments can create big leverage. It really feels like we’re standing at a new turning point. 🌟 We’re not being replaced. We’re being upgraded — if we choose to be. And for designers who lean in, the next few years might be the most exciting window of opportunity we’ll see 🙌 #productdesign #designer #genai #career #ai #careerdevelopment

Are you still designing a product interface or are you designing a human–AI ecosystem? We’re used to thinking of AI as a single feature inside a product. But Agentic AI shifts that entirely. We’re no longer designing for an AI model, we’re designing around a network of AI agents working together, planning tasks, and acting with autonomy. In a recent research paper, this was framed not as a UX challenge, but as a 'coordination design problem'. For designers, that raises new questions like: 1. How do we expose the system’s “thinking” without overwhelming the user? 2. When do we let the AI handle tasks silently, and when do we make its process visible for user oversight? 3. How do we avoid creating a black-box experience, especially when multiple AI agents are coordinating in the background? Designing for Agentic AI moves us beyond interfaces. It forces us to design trust, transparency, and control into the system architecture itself. Curious how others are thinking about human–AI coordination in their products? It’s a shift I’m starting to explore more seriously. I’ll be sharing everything I learn along the way, from frameworks to real world applications. Follow along if you’re curious about designing for the next generation of products too :)

The latest State of AI in Design report from Foundation Capital reveals something important -- AI adoption by designers is overwhelming, with many creating AI-powered toolkits across their workflow -- but most are feeling some gaps: ➡️ Still missing enhanced UI/UX generation ➡️ Design System isn't integrated ➡️ Missing Integrated Workflows ➡️ Need for more advanced prototyping capabilities The issue: Teams have the tools—ChatGPT, Cursor, Figma—but they're working in isolation. The Real Problem Isn't More Tools. While the vibe-coding tools are a game-changer, without proper systems, they have zero understanding of your product's purpose or behavior. Your design systems weren't built for AI consumption, which means: ❌ AI outputs require extensive human intervention ❌ Context gets lost at every tool handoff ❌ Integration costs multiply with each new AI tool ❌ Teams spend more time managing tools than benefiting from them At Superfriendly, we architect design systems that AI can understand and work with effectively. When your systems are properly structured, you unlock: 🎯 Contextual Design Generation AI that creates interfaces based on your specific brand constraints, design system rules, and user context—not generic outputs that need extensive modification. ⚡ Automated Design-to-Dev Pipeline Streamlined handoffs through automated code generation and component mapping that understands your tech stack and development workflows. 📚 Intelligent Documentation & QA Systems that auto-generate documentation from design files while tracking consistency and enforcing standards across your organization. 🧠 Knowledge Discovery & Assistance AI assistants that provide role-aware responses to designers, PMs, and developers, surfacing insights through intelligent search of your systems and best practices. 🔍 Proactive System Monitoring Automated tracking of design consistency with pattern matching that identifies quality issues before they impact user experience. The Strategic Window We're at a critical moment. Organizations continuing to accumulate point solutions will find themselves managing increasingly complex integrations. Those investing in AI-ready system architecture will build sustainable competitive advantages. Three immediate priorities for design leaders: 👉 Audit your AI integration overhead - Calculate the true cost beyond licensing fees 👉 Assess your system architecture readiness - Can AI actually understand and use your design systems? 👉 Invest in AI-native infrastructure - Address integration challenges rather than adding more tools The teams making these infrastructure decisions now will define the standards others follow later. Your next design system user won't be human. Is your system ready?

Design just entered a new era. And Figma Make is a big reason why. Figma + AI isn’t replacing designers. It’s removing friction. The boring parts. The slow parts. The “staring at a blank canvas” parts. What changes with Figma Make: → Faster iterations → Smarter components → Less gap between idea → execution → More time thinking, less time repeating For designers, founders, and product teams, this means one thing: Focus on outcomes. Not busywork. I’ve been experimenting with Figma Make and built a small proof of concept. An Expense Tracking App. Designed around the reality of an average person in Argentina. Fully responsive. Built for real life. What I focused on: → Exchange rate / currency conversion → Category-based tracking to understand spending habits → Configurable starting budget for a realistic baseline → Track both income and expenses to see full cashflow → Monthly transaction structure for clarity → Data export for deeper analysis → Built with Supabase integration in mind for future scale The goal wasn’t perfection. It was speed and clarity. Because this is the real product-design bottleneck: PM drops a massive PRD → designer stares at it → first prototype takes days By then, momentum is gone. I felt this in my own workflow. Too much time wasted just getting to the first draft. So I changed the process. Here’s what works now: → Take the PRD → Use ChatGPT to turn it into a Figma Make prompt → Paste it into Figma Make → Get draft screens the same day Now there’s something to react to. Refine. Discuss. Improve. Ideation, accelerated. A few lessons learned: → Start with one user flow → Write prompts as user stories (“As a user, I want to...”) → Never treat the AI output as final — it’s always a starting point This doesn’t kill creativity. It kills the blank canvas. And gives designers more time for what actually matters. What’s your take on AI inside design tools? Hype or real leverage? #FigmaPartner

🚀 𝐇𝐨𝐰 𝐀𝐈 𝐢𝐬 𝐜𝐡𝐚𝐧𝐠𝐢𝐧𝐠 𝐔𝐗 𝐝𝐞𝐬𝐢𝐠𝐧 — 𝐀 𝐩𝐞𝐫𝐬𝐨𝐧𝐚𝐥 𝐥𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐜𝐮𝐫𝐯𝐞 Six months ago, I opened Figma like I always did. Blank screen. Cursor blinking. The process was familiar: → Research → Wireframes → Prototypes → Endless tweaking Then I tried something new. I prompted ChatGPT: “Design a homepage for a health-tech startup.” 10 seconds later: → Suggested layout → Smart user flow → CTA ideas It wasn’t perfect. But it gave me a head start. Since then, I’ve changed how I work. 🧠 I use AI to explore structure ✍️ Draft copy 📊 Analyze user behavior ⚡️ And test ideas before building I don’t design less. I design with clarity. Faster. Sharper. With more confidence. But here's the truth: AI won’t replace great designers. It just replaces the slow start. Empathy, intuition, and real-world insight? Still 100% human. If you’re not integrating AI into your workflow... You’re not falling behind. You’re already there. Curious how other UX designers are using AI? Let’s swap notes. Drop a 💬 below. #UXDesign #AIDesign #ProductDesign #DesignThinking #ChatGPT #OpenAI #Figma #HumanCenteredDesign #TechTrends #DigitalProductDesign #PromptEngineering #FutureOfWork #DesignTools #JustinWelshStyle

Figma just published a paper on design systems and AI. Get your copy here → https://lnkd.in/gKSqgHru Thanks David Serrault for sharing the link 🙏 Design systems in the age of AI: strong foundation, but we need to go further This paper makes an important point: design systems are no longer just for humans. AI is becoming a first-class consumer of design systems — and that changes everything. But I think we’re still not pushing this idea far enough. If design systems are used by machines, they must be machine-first readable. Tokens, structure, naming, metadata, and documentation are no longer “best practices” — they are the interface. Any ambiguity humans can resolve intuitively becomes a failure for AI. Documentation also needs a shift. Explanations are not enough. We need explicit, practical rules: - what must be used - what must not be used - under which conditions AI doesn’t need inspiration. __ It needs constraints and directives. __ 🦾 💥 Another critical point: system quality becomes non-negotiable. 💥 Machines can’t reliably self-correct yet. If the design system is wrong or inconsistent, AI will amplify those flaws at scale. Design systems now need to be tested in generative-UI contexts, not only in static design or production UI. Finally, leanness matters more than ever. 🚀 A lean system — fewer components, fewer variants, clearer intent — is easier for AI to reason about and harder to misuse. Bloat is no longer neutral; it’s a liability. Why this matters today: 😱 AI is already producing UI faster than teams can review it. 🤩 Design systems are becoming the last line of defense between coherent products and generic, degraded interfaces. Getting this right now determines whether AI scales design quality — or quietly destroys it. That’s where a ` lean system ` with simple, clear, and smart rules can make the difference. #DesignSystems #AIinDesign #GenerativeUI #UXDesign #ProductDesign #DesignOps #DesignInfrastructure #Figma #AIProduct Friends of Figma, New York City, Friends of Figma, Austin Alexia Danton