🤿 slow is smooth, smooth is fast

I’m willing to bet that you have been on Smashing Magazine at some point in the last 18 years…

Their founder, Vitaly Friedman, has been one of the leading thinkers in UX for decades. And right now he’s obsessing over how we can design AI experiences that people love using.

So this week’s episode is a masterclass in design patterns for AI (read: lots of screen-sharing 👀).

We dissect products like Consensus, Perplexity, Elicit, and many more to figure out what’s missing and what can be improved.

Some highlights:

• The use case for dynamic interfaces with AI

• How to design a less painful refinement journey

• The best AI design patterns to use for inspiration

• When to use quiet AI vs. visible AI in your interfaces

• Why more products should be “AI-second” not “AI-first”

• Why we need to slow users down when designing AI products

• How designers can establish trust when users interact with AI

• + a lot more

Listen on YouTube, Spotify, Apple, or wherever you get your podcasts 👇

🤝 WITH PAPER

Remember what Janum Trivedi from Airbnb said about shaders?

They used to be this alien format that I couldn’t really do anything about. But now Paper is making it possible for designers anywhere to create their own shaders.

You can even preview them directly on the canvas…

It’s just another reason why I am all-in on Paper as the next great design tool.

They’re truly raising the ceiling for creativity and you can be one of the first to try it out just click the link to get started 👇

🔑 KEY TAKEAWAYS

Key takeaways about AI design patterns

1 — Helping users slow down

I worked with a contractor on my home for a year and he would always say the same phrase: “Slow is smooth, and smooth is fast”.

I think this applies to designing AI products as well 😅

Vitaly’s goal is to make sure that before anybody sends a prompt, it’s “so succinct, so accurate, so useful, so detailed, so contextual” that the chance of getting a generic, unhelpful response is minimal.

That means actively slowing people down before they attempt to generate an output with AI.

Skilled prompters would never open a new chat and say “make me a ______”.

They would probably say something like: “My goal is to make a _______. I’m going to include some context, and then I’d like you to ask me as many questions as needed to ensure you have everything required to nail the task”.

So how do we bake that understanding into the UX itself?

The good news is we already have some tried and true patterns for doing exactly that👇

2 — Leveraging familiar UI patterns

The internet is a sea of open-ended text boxes right now.

Proven UI patterns are strangely absent in AI interfaces, even though they’ve been table stakes in SaaS for years.

It’s a big reason why Vitaly is such a fan of Consensus. They layer familiar affordances around the core chat interface that really improve the UX (like a filter panel for instance).

They also go further with ideas like quantifiably representing sources in a “Consensus Meter” and creating UIs to filter by type. These patterns make results more useful and help users feel in control.

Vitaly shares about 20 more examples of little patterns like this in the full video 😅

3 — Dynamic loading states

Loading states are more important than ever because this is the first time it actually makes sense to have a 10–20 second wait in your product.

So how can we use that downtime to help users add context and set the model up for success?

It reminds me of something the Head of Design at Gamma shared: in the early days while presentation outlines were loading, they gave users theming options to customize the output. This eliminated the feeling of waiting and turned a dead moment into progress.

How many other AI products could take advantage of this same moment?

Based on the initial prompt, the AI could assemble a UI to clarify the user’s intent.

Maybe that includes sliders, a number rating scale, a follow-up question, etc. All of this could be done with simple, predictable components.

Done right, these micro-flows would keep people engaged and level up the quality of AI interactions.

So if you enjoy getting nerdy on patterns like this then this episode will probably be one of your favorites yet 👇

Meet the Dive partners 

I made a list of my favorite products and asked them to come on as sponsors of the newsletter/podcast. They said yes 🥹

The #1 way to support Dive Club is to check them out👇

Framer​ → How I build my websites

Genway → How I do research

Granola → How I take notes during CRIT

Jitter​ → How I animate my designs

Lovable → How I build my ideas in code

Mobbin → How I find design inspiration

Paper → How I design like a creative

Raycast ​ → How I stay in flow while I work

Thanks for reading! I'm working hard to bring you the best design resources on the planet 🫶

If you want to go even deeper you can always:

☞ Read all ​ideas​

☞ Watch all ​episodes ​

☞ Learn ​advanced design tactics​

P.S. if you were forwarded this email you can ​subscribe here