Escape from the Figma Titanic: Part 1 — Deep Dive into AI-Driven Use Cases

If you’re starting the design of your AI product by obsessing over pixel-perfect Figma screens... You’re on the wrong ship.

Designing for AI is not just about the UI. It’s about aligning the system’s intelligence with real human needs — and that starts with the right use case. Get it wrong, and no amount of beautiful screens will save you. Get it right, and your AI system becomes not just functional, but trustworthy, ethical, and actually helpful. (Not to mention you can basically print money… Which makes shareholders happy and helps UXers and the rest of your team stay employed.)

The best way to align AI with user needs? Storyboarding your use case.

Here are five essential DOs and DON’Ts when crafting storyboards for AI-powered systems, drawn from 35 AI-driven projects and decades of field experience distilled from the UXforAI.com series on Storyboarding for AI: https://www.uxforai.com/p/storyboarding-for-ai-part-4

✅ Top 5 DOs for AI Storyboarding

1. DO start with user intent, not the algorithm.
A good AI use case storyboard begins where the user begins — with a motivation, a goal, a pain point. What is the human trying to do? What friction exists today? Your AI is supporting and assisting the user – it is a helper, and the user is the star. If you don’t root the storyboard in the user’s reality, you’re designing magic tricks, not meaningful tools.

2. DO test your storyboard before touching Figma.
Storyboards aren’t just alignment tools — they’re low-cost user research and validation artifacts. Review and validate them with users! 

Ask questions like: 

• Do you do this today? If so, how? Can you show us?

• What are your current pain/friction points with the manual process?

• If you were King/Queen for a day, what would the ideal process be like?

Make the changes right there on the call – this is why we use square stickies – you can move them around and replace individual panels easily in seconds! For extra awesomeness, show your sketching work live using a document camera. Keep asking:

• Is this what you mean? (If not, iterate again until you get it right.)

Then ask your million-dollar question:

• How much would you pay for this? 

If you did this right, the response should be enthusiastic. Your customers should be saying unprompted, “When can I get this?” If they do not – this is not the right AI-driven use case. Pivot and move on.

Congratulations, you’ve just saved weeks of misguided Figma robot monkey work.

3. DO revise your storyboard multiple times. Together with your team.
Recently, I have come across this yet again. Just like hundreds of times in the past, the idea that very smart expert people in the room came up with was not quite right. As soon as we tested the storyboard with customers, it became clear that the Agentic AI flow we were so excited about needed to be tweaked. Is the new flow something that AI can handle? You don’t need to answer every question like that by yourself – you have a 4-in-a-box team. UX is the crazy glue, the catalyst, the amplifying factor – this is why you should make it a point to connect with your PM, Dev lead, and Data Scientist at least on a weekly basis, report your findings and iterate on the flow together.

4. DO use Subject-to-AI transitions.
This is the most powerful pattern for AI UX: The user takes an action... and the AI steps in with a helpful contribution. For example:
👩 “User sends a voice message” → 🤖 “AI transcribes and summarizes.”
These transitions clarify when, where, and why the AI activates. Without this awareness, you’re just back to drawing Figma screens.

5. DO consider what the AI sees, knows, and does.
Don’t just draw what the user sees. Consider system awareness: inputs like location, text history, sensor data — and outputs like decisions, predictions, or suggestions. Ideally, your story should communicate what the AI perceives and outputs indirectly, by walking through the user experience. This makes the invisible visible.

Bonus Tip: DO explore edge cases, errors, and escalation paths.
The happy path is only half the story. Where might the AI misfire? How do you gracefully fail? Where does a human take over? Exploring and sketching these branches with your customers helps you design with safety, ethics, and transparency in mind.

❌ Top 5 DON'Ts That Sink AI Use Cases

1. DON’T storyboard system features — storyboard human outcomes.
“AI sorts documents by category” is not a use case. “User finds what they need 10x faster” is. Don’t make the AI the hero. Make the user’s outcome the narrative arc.

2. DON’T skip the AI handoff moment.
Many storyboards gloss over the moment of trust: when the AI acts on behalf of the user. This is a massive UX moment — did the user ask for it? Can they see why the system acted? If your storyboard skips this, you’re designing a black box.

3. DON’T use static wireframes to explain dynamic intelligence.
Figma is great — for static UI. But AI isn’t static. It adapts. It reacts. If you rely solely on screen mockups, you’re flattening the intelligence out of the experience and only considering the visible part of the iceberg. Storyboards are how you can model AI processes across time and distance.

4. DON’T design in isolation from data science and engineering.
You don’t need to know TensorFlow, but you do need to know enough to ask the right questions and sanity-check your AI-driven UX ideas. Many flawed storyboards assume the AI will “just know” what to do. Collaborate early. Learn the capabilities of your team and lean on them for help! Your data science team is your co-creator. 

5. DON’T assume your first storyboard is the right one.
Designers are used to iterating on UIs. Iterate on use cases, too. The first version might be shallow, unrealistic, or irrelevant. Refine it. Run it by users. Let it evolve. It’s your map — not your masterpiece.

📘 Want the Full Blueprint?

I consider use case identification so important that I devoted the entire Part 1 of my new UX for AI book to “Framing the Problem.”

And I wanted you to own the complete blueprint for UX-for-AI research and design process — from choosing the right use case to crafting ethical, human-centered AI systems.

This is why I wrote UX for AI: A Framework for Designing AI-Driven Products.

My 6th UX book is the #1 new release in Data Modeling and Design on Amazon and features insights from 12 co-authors — UX luminaries like Jakob Nielsen and Josh Clark. Together, we created this book as a guide to help you escape from the Figma Titanic.

Don’t wait for the ship to sink.
Get your copy here → https://amzn.to/4l2ShyL 

Till next time,
Greg