Hey friend,

This is Design Current. A newsletter about user experience design.

Lately, I’ve felt a bit… gaslit by AI products about my prompts.

Any time the output misses the mark, I’m told:

But why should users know what is a good prompt???

This is a real issue for AI products that rely on chat interfaces.

As designers, we take this seriously.

This is what we did to improve AI chat UX and, in turn, boosted product adoption.

1. We reduced uncertainty by showing result confidence

What we noticed:

Interacting with a blank chat box was revealing.

Many people look at the input field, unsure where to even begin.

Others typed something short, got a weak output, and immediately assumed the product wasn’t very capable.

What we created:

When redesigning the AI experience for Quiller.ai (a content creation AI product for mission-driven organizations) we introduced a new pattern: output confidence badges.

The product shared how confident it felt about the response, so users could quickly judge whether their prompt needed more detail.

Why it worked:

It gave users a clearer sense of how the AI was interpreting their input and if there's potential to getting better outputs.

With expectations set, people felt more comfortable exploring and iterating.

2. We guided users with clarifying questions

What we noticed:

When we listened to user sessions, we kept hearing the same hesitation:

“Wait… what exactly am I supposed to tell it?”

People wanted to give good input, but they weren’t sure what the AI needed. The result was disappointing outputs, which only reinforced the idea that the product wasn’t helpful.

What we created:

We added friendly, targeted clarifying questions inside the chat flow. These questions helped users clarify their goal and supply the right details.

​
​Why it worked:

Users no longer felt like they had to guess. Their prompts improved naturally, and so did the outputs. When people see better results, they come back.

3. We used structured input when chat wasn’t enough

Chat is great for conversation, but not always for precision.

In some workflows, we found it was better to move away from chat entirely, and return to a well-structured form-like page.

When users know what they want and need to enter several pieces of information, a traditional form can make the process easier.

It provides structure, clarity, and confidence—often more than a freeform chat can.

TL;DR

AI chat products can make big promises, but if users cannot quickly understand the value, they will not stick around.

These patterns made a meaningful difference for us:

• Show result confidence to set expectations
• Ask clarifying questions to guide input
• Use structured forms instead when a task needs precision

Until next time,

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​