Hi friend,

This is Design Current, your go-to newsletter about user experience design, brought to you by Koi Studios.

Studio Update Applying UX to democracy?! I (Anyi) led a journey mapping session at Harvard Ash Center for Democratic Governance and Innovation. Alongside political scientists, campaign staff, election administrators, technologists, and philanthropists, we mapped the end-to-end journey of American voters. ​Read more about it here. More to come, stay tuned.

---

4 AI chat UX patterns we're loving right now

Chat is still the dominant interface for AI products.

The irony with a chat interface is, although the act of chatting itself is intuitive, a blank chat box can feel intimidating.

We think about this a lot, and we're always on the hunt for thoughtful ways to improve chat UX. Here are a few patterns we love.

1. In the moment tips

What it is: Bite-sized, contextual nudges that show up right when a user might need them. Not a tour, not a modal, not a checklist. Just a tiny hint at the right moment.

Why we like it: Onboarding is hard for AI products. The canvas is open, and people learn by doing. Small tips give users just enough handholding to get started, without overwhelming them.

And it doesn't have to stop at onboarding. The same pattern can help experienced users too. It can suggest shortcuts, or surface tips based on how their product usage.

2. Prompt templates

What it is: Pre-written prompts users can click to run instantly, usually surfaced on the empty state or near the input field.

Why we like it: Typing out a long, well-crafted prompt can be tiring. Templates solve this nicely.

For new users, they lower the activation energy to get a first result, and they teach users what the product is actually good at.

For existing users, they're a shortcut. Once someone knows what they want, a one-click template is a much faster path.

3. Customizable outputs

What it is: Letting users define how they want the AI to respond. Tone, format, structure, level of detail.

For example, A notetaking app might let you choose bullet points over prose, or ask the AI to always bold key decisions and attribute quotes to speakers.

Why we like it: After using a product a few times, people start to develop preferences. A product that remembers those preferences feels a lot more personal. It also changes the relationship from "AI gives me something" to "AI works the way I work," and that's a much harder thing to walk away from.

4. Visual emphasis on reasoning

What it is: Showing the steps, sources, or logic behind an AI's answer, often through expandable sections, inline citations, or a visible chain of thought.

Why we like it: When users can see how a model got to its answer, two good things happen. They're more likely to believe the output, and they're better equipped to course-correct when something's off. Reasoning transparency turns AI from a black box into a collaborator.

The through-line

If there's one thread connecting these four patterns, it's giving users a sense of being in control. Small tips, ready-made prompts, customizable outputs, and visible reasoning all give people a way to shape and understand what's happening in the chat.

That sense of control can really influence how much users trust and like a product.

What do you think about these patterns? You can always reply directly to this email and chat with us.

Thanks!

Anyi & Andrea

Cofounders @ Koi Studios​

If someone forwarded this to you → join us​
If you're looking for product design help → chat with us

600 1st Ave, Ste 330 PMB 92768, Seattle, WA 98104-2246
​Unsubscribe · Preferences​