AI UX Design Patterns
Where to learn UX design for AI interactions
👋 Hello, it’s Niki here! Welcome to this ✨ free edition ✨ of The Product Microscope newsletter, where we deep-dive into the world of UX, sharing practical tips and insights that will take your product, UX design, and user research skills to the next level, enabling you to make a greater impact as a design practitioner and accelerate your career growth.
If you’re a product or UX designer right now, you’ve probably noticed that “design for AI” is no longer a niche specialty (as it was 2 years ago).
It’s just something that casually lands on your desk, whether you asked for it or not.
Maybe you’re adding a chatbot to an existing product, figuring out how to surface AI recommendations without creeping users out that AI is watching them, or just trying to understand what your engineering team means when they talk about “agentic workflows.”
I have good news for you!
You don’t have to figure this out from scratch.
Some of the biggest names in tech, such as Google, Microsoft, IBM, and SAP, have been quietly publishing frameworks, pattern libraries, and guidelines for designing AI experiences.
It’s even better that these resources are free and specifically written for people like us who need to design AI-powered product features.
I went down a rabbit hole exploring what’s out there, and I’ve pulled together the resources that are actually worth your time.
Whether you’re brand new to AI design or looking to level up your toolkit, these will give you AI design patterns, practical real-world examples, and the vocabulary to have more intelligent conversations with your cross-functional partners.
Watch the video version of this article on YouTube:
Resources for Digital Product Design
Carbon Design System: Carbon for AI
IBM’s Carbon for AI is an extension of the Carbon Design System specifically for identifying AI-generated content and delivering explainability in products.
What’s unique about this is that they use light as a visual metaphor (brightness, glow, and gradients) to distinguish AI-generated or AI-recommended content from standard UI elements.
The framework provides styling guidelines, AI-specific tokens, and component patterns to help build user trust by being transparent about when and how AI is involved in an experience.
Google PAIR Guidebook: User Needs + Defining Success
The People + AI Research (PAIR) team’s guidebook could be used as a foundational resource for building human-centred AI products.
This specific chapter focuses on identifying user needs, determining when AI is the right solution, and designing reward functions that optimize for long-term user benefit.
It includes worksheets and exercises for problem framing, and has been updated for the generative AI era with guidance on handling hallucinations, fact-checking, and appropriate UI for explaining how AI works.
The Human-AI eXperience (HAX) Toolkit is Microsoft’s suite of practical tools for creating responsible human-AI experiences.
It includes 18 research-validated guidelines organized around four interaction phases: initially, during interaction, when AI is wrong, and over time.
The toolkit helps you conceptualize AI system behaviour early in the design process, with practical resources such as the HAX Workbook for planning and the HAX Playbook for anticipating failure scenarios.
Microsoft HAX Toolkit: AI Guidelines
These guidelines synthesize 20+ years of human-AI interaction research into actionable best practices.
Each guideline includes examples and design patterns for implementation, with the Design Library offering searchable implementation examples filtered by guideline, product category, and application type.
The guidelines address fundamental questions like “How do we communicate what the system can do?” and “How do we help users correct AI mistakes?”
Also, a video comes with it with an explanation - even though it’s over 4 years old, it’s still relevant.
Microsoft Design: UX Design for Agents
This article is only a 10-minute read, but I’d highly recommend it if you want to learn about design principles for AI agents and multi-agent systems.
The principles span three categories:
Space (how agents engage in physical/digital worlds)
Time (how agents use memory and context across past, present, future), and
Human-Agent Interaction (transparency, control, trust)
The framework can help you design agents that connect people to knowledge while maintaining appropriate user oversight and privacy.
Shape of AI is a UX design pattern library specifically for AI-powered interfaces, curated by Emily Campbell.
It catalogs patterns across the entire AI interaction lifecycle:
Getting started (onboarding, sample generations)
Prompting (contextual input, clarifying questions)
Output handling (regeneration, summarization), and
Trust-building (transparency, human-in-the-loop controls)
This has, very quickly, become my favourite resource for understanding AI interface conventions.
SAP Fiori: Designing Intelligent Systems
SAP’s guidelines for AI in enterprise software focus on two pillars:
AI automation (freeing users from repetitive tasks), and
AI augmentation (guiding users to better decisions)
The framework emphasizes finding the “sweet spot” where AI capabilities overlap with user needs, and it covers UX design patterns for notifications, recommendations, matching, and situation handling.
It also includes guidance on proactive vs. reactive AI assistance and different automation levels.
This SAP guideline addresses how to communicate AI decisions to users to build trust.
It covers progressive disclosure of explanations, confidence indicators, and how to present algorithmic reasoning in business contexts where incorrect recommendations can cause real damage.
The guidance emphasizes that explainability isn’t just ethical; it’s essential for user empowerment in high-stakes enterprise decisions.
Google Cloud: 101 Real-World Gen AI Use Cases with Technical Blueprints
Google Cloud published 101 architectural blueprints as practical starting points for AI projects, covering domains such as retail, healthcare, and finance.
Each blueprint shows a design pattern with a corresponding tech stack to solve real challenges, from automating document summarization to preventing fraud.
It’s a technical complement to their 1000+ customer use cases, offering UXers concrete examples of how AI is being implemented in products across industries.
Where To Start?
The landscape of AI design is moving fast, but the fundamentals of good UX, understanding user needs, building trust, designing for failure, and keeping humans in control haven’t changed.
What’s shifted is how we apply those principles to systems that are probabilistic, adaptive, and sometimes unpredictable.
These resources won’t give you all the answers (nobody has those yet), but they’ll give you a solid foundation and a shared language to work from.
My suggestion?
Start with whichever resource matches your most immediate challenge, whether that’s explaining AI decisions to users, designing your first agent interaction, or just getting your team aligned on what “good” looks like when it comes to designing AI-powered product features.
And if you find other resources that belong on this list, I’d love to hear about them. We’re all learning together.
And…A Note on AI Design Tools
Personally, I’d rather spend my time levelling up my actual design craft than chasing every new AI tool that launches.
AI tools are important, sure. But they’re also temporary.
I started designing in Adobe Fireworks. Anyone remember that?
Then it was Photoshop, Illustrator, Axure, Sketch, XD, Figma, Framer...and now Lovable, Replit, and whatever launches next week. The design tool graveyard is so real.
But what haven’t changed are:
The fundamentals
Understanding users
Creating clear systems
Simplifying complexity
Strategic thinking
Collaboration
These skills work in any tool, and they’re what actually make you a principal-level designer.
That’s exactly why I’m drawn to resources like the ones above.
They’re not teaching you how to use the hottest new AI design tool. They’re teaching you:
How to think about designing AI experiences
How to build trust
How to design for failure
How to keep humans in control
How to explain algorithmic decisions
These are craft skills, not tool skills.
Whether you’re designing in Figma today or whatever replaces it tomorrow, understanding how to design transparent AI systems will still matter.
The frameworks & patterns from Google, Microsoft, and IBM aren’t tied to any specific tool; they’re tied to human behaviour and sound design principles.
So that’s where my focus is; and that’s where your focus should be too!
Craft over tools.
The tools will keep changing. The craft is what lasts.
Brilliant. Highlighting the 'agentic workflows' communication gap is crucial for interdisciplinary team.