# Design for the moment the AI is unsure.

> The design of an AI interface is not complete when it shows a good answer. Users need to understand what the system knows, what it did, what it has not checked and what they can safely do next.

Canonical URL: https://marmation.com/solutions/ai-ux-and-product-design/

## The decision

Can customers understand, correct and reject the suggestion?

Show the source, the proposed change and anything still unknown before asking for approval. Choose a form, comparison or work queue when that makes the task clearer than a conversation.

## A suggested record update still needs a human decision

An AI drafting feature proposes an update to a customer record. The user must decide whether the result is based on the correct source, what changed from the original and whether to approve the write.

## A chat window can hide what has and has not been checked

A chat-first interface can hide the structure of the task. Generic confidence badges often do not help. Users may accept fluent text as certainty even when the application lacks evidence.

## Make changes, evidence and recovery visible

Design what people see when information is missing, a suggestion is wrong or an action fails. Show where the evidence came from, how to correct or reject a change and whether undo is actually supported. These states must match the backend: a reassuring button cannot reverse a change unless the system can do so. Test the ordinary task and its recovery path together.

## The review and correction patterns the product needs

- Task flow and interface states for AI and non-AI paths.
- Cited evidence, compare/review, edit and escalation patterns.
- Design system components for loading, refusal, uncertainty, audit and undo.
- Usability tests paired with model behavior and error conditions.


## A table or comparison may serve the task better than chat

Not every AI interaction should be a conversation. A structured review table, command panel or side-by-side comparison may be more useful than another chatbot.

## Can users spot a bad source and understand when a change is final?

Test whether users can recognize incorrect evidence, inspect source context, recover from bad output and understand when an action becomes final. Accessibility must work with dynamic content.

## Design the error and review journey first

Start by designing the error and review states next to the best-case journey. That is where a product earns trust.

## Related reading

- [Build AI into the product. Not just the prototype.](https://marmation.com/solutions/ai-product-engineering/)
- [The best AI interface is honest about uncertainty.](https://marmation.com/review/designing-ai-ux-for-uncertainty/)
- [Give an agent a job. Give the system boundaries.](https://marmation.com/solutions/ai-agents-and-workflows/)