The model behind your chatbot might be world-class, but users only experience the interface. A blank text box that says "Ask me anything" is the most common — and most punishing — way to ship a chatbot. This guide covers the conversational UI patterns we use when building AI chatbots for businesses across the USA, UK, Canada, Europe and South Africa.
Conversational UI is more than a chat thread
The best conversational interfaces blend natural-language chat with supporting UI: suggested prompts, quick-reply chips, rich cards (an order, a product, a booking), inline citations, and a clear, always-available input. Pure free-text puts all the cognitive load on the user; well-placed UI affordances reduce effort and steer people toward what the bot does well.
Solve the blank-box problem with onboarding
Open with one short message that says what the bot can do, then show three or four tappable example prompts. This is the single highest-impact pattern in chatbot UX: it demonstrates capability, removes the "what do I even type?" hesitation, and gets users to a successful first answer fast.
Design the message, not just the text
Keep responses scannable — short paragraphs, bullets where helpful, and bold for the key takeaway. Stream tokens so the reply appears immediately and feels alive. Use a typing indicator for any pause. For structured answers (a price, a status, a list of options), render a card or buttons instead of a wall of prose.
Give users low-effort ways to continue
After most replies, offer the obvious next steps as chips: "see more", "book a call", "talk to a human", "start over". Cheap continuation keeps the conversation moving and turns a single answer into a completed task. Always keep free-text input available for anything off-script.
Fail gracefully — and hand off well
When the bot can't help, it should say so plainly, never invent an answer, and offer a way forward: rephrase, a relevant link, or escalation to a human. A timely, context-carrying handoff to live support is a feature. Pass the conversation history along so the user doesn't have to repeat themselves.
Build trust into the conversation
Be transparent that it's AI. Keep a consistent persona and tone that matches your brand. Show sources for factual claims. Confirm before any real action ("Shall I book the 3pm slot?"). These are the same trust principles from our AI product UX guide, applied to the chat surface.
Mobile, accessibility and performance
Most chat happens on phones: comfortable tap targets (44px+), an input that doesn't get hidden by the keyboard, and a thread that scrolls to the newest message. Support keyboard navigation and screen readers, announce new messages politely for assistive tech, respect reduced-motion, and keep the UI responsive even while the model is "thinking".
Frequently Asked Questions
What is conversational UI?
An interface where users interact through natural-language messages rather than forms — usually a chatbot. Good conversational UI blends the thread with suggested prompts, quick replies, rich cards and citations.
How do you onboard users to an AI chatbot?
Open with a short capability message and 3-4 tappable suggested prompts. This solves the blank-input problem and gets users to a successful first interaction.
Should an AI chatbot offer buttons as well as a text box?
Yes — mixing suggested prompts, quick-reply chips and rich cards with free text reduces effort and guides users toward what the bot does well.
How should a chatbot handle when it can't help?
Acknowledge it can't answer, don't fabricate, and offer a path forward — rephrase, a link, or a clean handoff to a human with the conversation context.
How do you make a chatbot feel trustworthy?
Be transparent that it's AI, keep a consistent persona, show sources, stream replies, never fabricate, and confirm before any real action.
Build a chatbot people actually like using
We design and build AI chatbots with conversational UX done right. Book a free 30-minute strategy call.