Designing AI-Native Products in 2025
I’ve spent the last eight years designing B2B SaaS products, and lately, I’ve found myself at the forefront of what I believe is the biggest shift in our discipline since mobile-first design: AI-native experiences. The traditional approach, such as static interfaces and purely manual user flows, isn’t enough anymore. AI is transforming how we design and what we design, pushing us to adopt new patterns like adaptive UIs, conversational interfaces, and autonomous agents that change user expectations overnight.
In this post, I’ll share some personal insights on designing AI-native products, drawing on my work with AI agents and the broader research showing how predictive, personalized, and automated interfaces rapidly redefine the UX world. I’ll also highlight the new skills designers need in 2025 because if there’s one thing I’ve learned, it’s that the pace of AI-driven innovation means we have to learn faster than ever before.
My Journey into AI-Native Design
For most of my career, I worked on typical B2B SaaS software: think multi-step workflows, dashboards, and feature sets catering to niche enterprise needs. But in the last couple of years, I’ve been immersed in building AI-native products, tools where AI isn’t just an add-on but the core of the experience.
Right now, I’m working on an AI agents project that blends user interviews with domain-expert knowledge to create “productized” experiences. These agents capture the unique “taste,” decision-making processes, and methodologies of experts in fields like marketing and advertising to transform them into automated workflows. Instead of designing screens that require user input at every step, we’re training AI models to take on complex tasks, proactively suggest next steps, and adapt to each user’s context, all while maintaining transparency and trust.
This kind of work is a whole new ballgame. It merges user-research best practices (like extracting a subject-matter expert’s hidden knowledge) with advanced AI capabilities (like real-time learning or generative text). The result is an interface that feels more like a conversation than a static set of forms.
Why AI-First is Reshaping UX
AI dramatically enhances user experience by making products more predictive, personalized, and automated.
Predictive: Machine learning models can analyse user behaviour and data to anticipate what someone needs before realising it. In my AI agents project, for example, if an expert always checks a specific data point before making a decision, the agent learns to surface that data at precisely the right time.
Personalized: We’ve all experienced personalized playlists on Spotify or product recommendations on Amazon. Now, every SaaS vertical wants to offer similarly hyper-personalized experiences where AI can dynamically shape the UI to each user’s needs and preferences.
Automated: From resizing images to running A/B tests, AI can free designers from repetitive tasks. Meanwhile, end-users enjoy features like chatbots, smart defaults, and predictive text that reduce friction. Automation is especially powerful in enterprise software, where forms and processes can be mind-numbingly long. An AI that auto-fills fields or reorganizes layouts saves enormous time and headaches.
New Design Patterns for 2025
As we step further into 2025, designers need to develop new skills and patterns that centre on AI’s unique capabilities. Here are a few I see emerging:
Adaptive UI or ‘AUIs’
Traditional interfaces are static, but in an AI-native product, the UI might adapt based on context. Instead of giving every user the same dropdown menus and fields, we can use an AI model to anticipate which component is relevant. For instance, in a recent project we enriched conversational interfaces with enriched components to create a more engaging and personalised experience that goes beyond just text outputs. Designing these adaptive components requires thinking about states, triggers, and fail-safes, so the interface never feels out of control or opaque.
Amplitudes ‘Text to UI’
Conversation Design
Chatbot UIs and voice assistants have moved beyond simple FAQ bots. They now handle complex tasks, from scheduling to troubleshooting software. Conversation design involves writing dialogue, defining the AI’s “personality,” and planning fallback flows for when the bot inevitably stumbles. This requires new skill sets, part copywriter, part UX researcher, part interaction designer. Crafting a natural language experience that also delivers business value is challenging but key in AI-native products.
Sana AI's conversational chat
Data-Driven Iteration
AI-driven products can learn from user interactions and update themselves. That’s fantastic for personalization but can cause confusion if the interface randomly changes. Designers must create guardrails and consistent patterns to let users know why the product is adapting and how they can control it. Iteration becomes continuous. Less about major release cycles and more about incremental improvements to the AI model’s performance.
Invisible AI
Invisible AI seamlessly integrates machine intelligence into a product’s flow, requiring minimal user intervention or obvious “AI” labeling. When done well, it can automatically surface contextually relevant information, streamline tasks in the background, and free users from tedious steps—resulting in an experience that feels magically intuitive rather than overtly computational.
Ambient AI
Ambient AI extends beyond the screen to sense and adapt to the user’s environment. It’s constantly aware of context like location, time, or real-time activity. This allows the system to proactively offer assistance, adjust settings, or trigger workflows without the user explicitly asking, creating an experience that is quietly helpful and deeply personalized.
The Fast-Changing Pace of Product Development
One of the biggest shifts I’m seeing is the pace of building. Tools like Replit, v0, and other AI-powered prototyping platforms let you spin up complex features or run feasibility tests in record time. You can instantly prototype conversation flows, generate UI variations, or even integrate a large language model (LLM) without heavy engineering.
Because development is speeding up, product strategy is becoming even more critical. If you can launch an MVP in days instead of weeks, the biggest challenge is no longer how to build something, but rather what to build. Designers must collaborate tightly with product managers and engineers to prioritize features that truly move the needle.
My Evolving AI-Enhanced Workflow
In addition to designing AI-native products, I’ve found that my own day-to-day work has been transformed by a powerful ecosystem of AI tools. Here are some of the ways I’m integrating them into my process:
CustomGPT & Claude: I rely heavily on a custom-trained GPT instance that understands my business context, user personas, and past user research. This “personal AI” helps with everything from brainstorming feature ideas to crafting UX copy. I probably interact with it hundreds of times a day—iterating on designs, refining product messaging, and even soliciting feedback on creative strategy.
Perplexity.ai: I can’t remember the last time I used Google search. Perplexity.ai’s natural language queries and summarized results save me hours, letting me bypass multiple tabs and get quick, curated insights in one place.
Napkin.ai & Galileo.ai: These are my new go-to design helpers. Napkin.ai applies visual frameworks to my rough ideas, helping me shape initial concepts more coherently. Meanwhile, Galileo.ai can propose rough layouts and UI components that is perfect for that first pass at a design when I’m short on time.
Replit & v0: For rapid prototyping I can spin up a simple AI-based app or concept in a fraction of the usual development time, test it with real data, and iterate quickly, often over the course of a single afternoon.
Agentic Workers: This is a massive multi-chain prompt library that integrates with GPT, streamlining any repetitive or large-scale prompt-based tasks. It’s like having an army of specialized AI agents ready to tackle different parts of the design or research process.
Beautiful.ai: Presentation decks are still a fact of life, and this tool automates much of the layout and formatting work, allowing me to focus on the story I’m telling rather than slide design minutiae.
Granola / Sana.ai & Read.ai: Organizing meeting notes, transcripts, and user interviews is easier with these tools, and then I can feed that data back into my custom GPT for deeper analysis. Read.ai even evaluates my presentation skills, suggesting ways to refine my delivery, helpful for stakeholder demos or pitching new features.
Serif.ai: Lastly, email management can be a slog, but Serif.ai helps automate inbox tasks, freeing me to spend more time on strategic design problems instead of responding to low-level queries.
These AI-driven tools don’t just save time; they fundamentally redefine my workflow, allowing me to focus on bigger-picture strategy and creativity. By letting AI handle the grunt work, whether it’s a rough layout, a quick user research summary, or a brainstorming prompt, I can devote my energy to the kind of high-level design thinking and experimentation that truly drives innovation.
Looking Ahead
AI-native design is still in its infancy, but it’s evolving rapidly. In just a couple of years, we’ve gone from seeing AI as a novelty add-on to a foundational software layer that personalizes UIs, automates tasks, and even shapes entire workflows on the fly.
For designers, it means branching into new territories: learning the language of AI, designing conversation flows, curating data sets, and building trust mechanisms. It also means refocusing on product strategy, deciding which problems are the most important to solve, and how to position your AI’s value proposition because the barrier to building is falling away fast.
I believe that by 2025, these skills and patterns will be must-haves. We’ll look back at our old static wireframes and realize how limiting they were compared to AI-driven adaptive experiences. Ultimately, the goal remains the same: create products that solve real user problems. But AI expands our toolkit so dramatically that we need fresh mindsets, processes, and ethics to reach its full potential.
I hope this post gives you a sense of how AI-native design is reshaping my own practice. If you’re interested in learning more or sharing your experiences, feel free to reach out!