آینده تجربه کاربری، رابط‌های بیشتر نیست؛ گفت‌وگوهای بهتر است.

For years, digital product design has been about arranging visible controls. We designed menus, forms, cards, filters, and dashboards so users could tell a system exactly what to do. Every action had a predefined path. If a user wanted something the interface had not anticipated, they had to translate their intention into the structure we provided.

AI is beginning to change that relationship. Instead of learning the interface first, users can increasingly describe an outcome in their own words. The important shift is not that every product will contain a chat box. It is that products are becoming capable of participating in a conversation about intent.

That sounds natural, but good conversational UX is much harder than placing an input field at the bottom of a screen.

A chat box is not automatically a conversation

I have seen products add a prompt field and call the experience conversational. The system accepts text, produces an answer, and visually resembles a messaging app. Yet it forgets earlier decisions, fails to clarify ambiguous requests, and gives users no reliable way to correct or inspect its work.

A real conversation has continuity. It recognizes context, asks useful questions, and lets both sides repair misunderstandings. When the stakes are high, it also makes uncertainty visible.

For a product designer, this means the core questions change. We are no longer only asking where a button belongs. We need to ask:

• What does the system believe the user wants?
• Which context is it using?
• When should it act, and when should it ask?
• How can a user review, edit, or reverse the result?
• What must never happen without explicit confirmation?

The quality of the experience depends on these decisions more than on the shape of the message bubble.

Intent is richer than a command

Traditional interfaces often force a complex intention into a sequence of small inputs. Consider planning a trip. A form may ask for dates, location, budget, number of travelers, and room type. A conversational system can let someone say, “I need a quiet three-day trip near Tehran, I will be working in the mornings, and I do not want to drive more than two hours.”

That sentence contains priorities and relationships that are difficult to capture in separate fields. But it also contains ambiguity. What does quiet mean? Does the two-hour limit apply to the full trip or each direction? Is reliable internet essential?

The system should not guess every missing detail. It should identify which unknowns could materially change the result and ask about those. Good conversational design is partly the art of asking the smallest useful question at the right time.

Memory must be visible and controllable

Memory can make an experience feel genuinely personal. If a system remembers my preferred tone, accessibility needs, or usual workflow, I can avoid repeating myself. But invisible memory also creates discomfort. Users need to know what is remembered, why it matters, and how to change or remove it.

I think of memory as a product surface, not a hidden technical feature. It needs its own states and controls. Temporary context should be distinguished from long-term preferences. Sensitive information should not quietly become permanent. When a remembered preference affects an answer, the system should make that relationship understandable.

The goal is not maximum memory. The goal is useful continuity with user control.

Trust comes from calibration, not confidence

Many AI experiences sound equally certain when they are right and when they are guessing. That is a design problem. A smooth response can make weak information feel reliable, especially when users cannot see the sources, assumptions, or actions behind it.

A trustworthy conversational product communicates the boundaries of its knowledge. It can say what it understood, surface important assumptions, and show a preview before taking a consequential action. It offers evidence when evidence matters. It lets the user correct the system without starting over.

This does not mean covering every response with warnings. Too much caution becomes noise. The challenge is to match the level of explanation and confirmation to the level of risk.

The interface will not disappear

I do not believe conversation will replace visual interfaces. Conversation is flexible, but visual UI is excellent for comparison, scanning, spatial understanding, and precise control. It is faster to compare four plans in a table than to hear them described one at a time. A calendar is often better than a paragraph for choosing dates.

The strongest products will combine both modes. Conversation helps users express intent and adjust direction. Visual interfaces help them inspect results, compare options, and make precise edits. One mode can hand work to the other without losing context.

For designers, this means designing a relationship between language and interface. A response might generate a visual plan. Editing the plan might update the conversation. The user should always understand what changed and why.

What product designers need to practice now

Designers still need typography, hierarchy, interaction patterns, research, and systems thinking. But conversational products add new material: prompts, repair strategies, uncertainty, model behavior, memory, and consent.

I find it useful to prototype conversations before screens. I write realistic user requests, including incomplete and frustrated ones. I map where the system should ask, answer, refuse, preview, or hand control back. Then I design the visual surfaces that make those moments clear.

The future of UX is not a blank screen where users must know the perfect prompt. It is a product that helps people express what they mean, shows how it interpreted them, and keeps them in control as intent becomes action. The winners will not be the products that talk the most. They will be the ones that listen, clarify, and respond with good judgment.