
Design AI experiences that are transparent, trustworthy, and keep users in control in just 40 minutes.
About This Template
What happens when your product's AI makes a recommendation that users disagree with or cannot confidently act upon?
The AI-Powered Interaction Prototyping Workshop helps product teams design how users interact with AI recommendations, correct mistakes, and respond to uncertain results. Using Miro's editable wireframes, UI components, and collaborative feedback tools, participants create three AI interaction states and evaluate how effectively each supports user understanding and control.
I designed this workshop for AI product managers, UI/UX designers, AI engineers, product strategists, and startup teams building intelligent software experiences.
Duration: 40 minutes Format: Individual or collaborative, virtual or in person Best For: AI product design, human-AI interaction, UX prototyping, AI transparency, and product development
How to Use This Template
Choose an AI-powered feature that provides recommendations or suggestions to users. Use the provided Miro frames, editable wireframes, and sticky notes to prototype three interaction states.
Step 1: Define the AI Interaction (5 minutes)
Miro Tools: Sticky Notes and Text
Identify the user's goal and the recommendation the AI will provide.
Document what the AI can suggest, what requires user approval, and how much control the user should have.
Focus on one clear AI-assisted task.
Step 2: Prototype the AI Recommendation (10 minutes)
Miro Tools: Wireframing Library, UI Components, and Text
Create a screen displaying an AI-generated recommendation.
Include a clearly labeled suggestion, a brief explanation, and controls for accepting, editing, or dismissing the recommendation.
Make it clear that the suggestion has not been applied until the user confirms it.
Step 3: Design Correction and Uncertainty States (15 minutes)
Miro Tools: Duplicate, Wireframing Library, Shapes, and Text
Duplicate the original screen to create two additional states.
Human Correction: Show how users modify an incorrect AI recommendation and confirm their changes.
AI Uncertain: Show how the system communicates uncertainty and provides an option for manual selection or review.
Keep the interactions consistent and the user's available actions clear.
Step 4: Test and Refine (10 minutes)
Miro Tools: Comments, Sticky Notes, and Frames
Ask participants to walk through three scenarios:
Accept a useful AI recommendation.
Correct an incorrect recommendation.
Complete the task when the AI cannot provide a reliable suggestion.
Observe whether users understand the recommendation, recognize uncertainty, and know how to proceed.
Document confusing interactions, missing controls, and opportunities for improvement.
AI Tip: Ask AI to review your three interaction states for unclear explanations, missing recovery options, unsupported confidence claims, and limited user control. Use its recommendations to guide revisions.
What You'll Achieve
Three editable low-fidelity AI interaction prototypes demonstrating recommendation, human correction, and uncertainty handling.
Use these prototypes to communicate AI behavior requirements, validate user interactions, align product and engineering teams, and prepare for usability testing.
Frequently Asked Questions
1. What is AI interaction prototyping?
AI interaction prototyping involves designing how users experience and respond to AI-generated suggestions, recommendations, or decisions before building the actual functionality.
2. How is this different from traditional UI prototyping?
AI-powered interfaces must account for uncertainty, incorrect outputs, and user oversight. The experience needs to communicate what the AI recommends and how users can respond.
3. Do I need AI engineering experience?
No. The workshop focuses on designing user interactions rather than building or training AI models.
4. Does the prototype need a working AI model?
No. Use simulated AI responses to illustrate different scenarios. The goal is to validate the interaction design before implementation.
5. Why create three interaction states?
Recommendation, correction, and uncertainty states help teams explore what happens when the AI provides useful, incorrect, or unreliable results.
6. How should AI recommendations be presented?
Label AI-generated suggestions clearly, provide useful explanations when available, and give users appropriate options to accept, modify, or dismiss them.
7. Should AI confidence scores appear in the interface?
Only when the scores have a validated meaning that users can interpret correctly. Avoid displaying arbitrary percentages that could create false confidence.
8. How should users correct an AI mistake?
Provide a straightforward way to edit or reject the recommendation. Make the result of their correction clear.
9. What happens when the AI cannot make a reliable recommendation?
Communicate the uncertainty and provide a suitable next step, such as manual selection or human review.
10. Can I use this template for different AI products?
Yes. The template works for AI recommendations, content suggestions, classification tools, intelligent assistants, and decision-support features.
11. Can AI help evaluate my prototype?
Yes. AI can suggest alternative interaction patterns, identify unclear explanations, and flag missing controls. Validate its recommendations through human review and usability testing.
12. Can I complete this workshop individually?
Yes. Create all three states independently, then share the prototypes with colleagues or potential users for feedback.
13. What should I do after completing the workshop?
Use the prototypes to document AI interaction requirements, define approval and correction behavior, and prepare usability tests before development.
Final Tip
Design AI experiences around user understanding and control. Good AI interactions communicate recommendations clearly, acknowledge uncertainty, and give users meaningful ways to review, correct, or reject the system's suggestions.
Deanne Watt
Product Strategy @ MiNDPOPGroup.com
My approach to product is to get to the heart of what drives a company. I am passionate about the entire end-to-end process and making it more efficient, collaborative as well as aligning teams and improving communication. We have built about 200 Miro boards so far that cover ideation, strategy, design, engineering, and even marketing promotion.
Categories
Similar templates


