The end of the chatbot? Designing AI interfaces that act and not just answer | EP15
The end of the chatbot? Designing AI interfaces that act, not just answer | Human × Intelligent EP15
Is the chatbox still the best default interface for AI?
In this episode of Human × Intelligent, Madalena Costa explores one of the biggest design shifts happening in AI product design right now: the possible end of the chatbox as the default interface for artificial intelligence.
This episode explores what happens when AI moves from answering to acting, why chat interfaces start to break inside real workflows, and what product designers, UX teams, founders and PMs need to learn now to design the next generation of AI experiences.
In this episode, you’ll learn:
Why chat won as the first dominant interface for AI
Why chat is still useful for some workflows, but not all
What changes when AI moves from conversation to action
Why many AI products do not remove friction and they just relocate it into language
The difference between a chat-first AI product and an embedded agentic experience
Why designers need to move from static screens to workflow-aware intelligence
The key UX patterns for agentic systems: previews, rationale, progress, undo and adjustable autonomy
When to use embedded AI patterns and when not to
How to redesign chatbot-style AI features into more contextual, trustworthy experiences
Why this matters
The future of AI interfaces is probably not invisible magic and not 'no UI'.
It is more likely:
embedded intelligence
contextual actions
proactive support
inline suggestions
smart previews
clear approval flows
strong trust and safety patterns
As AI systems become more capable, the interface does not disappear.
Its role changes.
The challenge is no longer just 'How do we let users prompt the model?'
It becomes: How do we help users express intent, understand action, review decisions and stay in control without doing everything manually?
Key ideas explored in this episode:
- Why chat won first: Chat is familiar, flexible, low-friction and easy to build. It made powerful AI accessible to millions of users.
- Why chat starts to break: Once AI systems can take actions, monitor progress, use tools and operate across workflows, forcing everything through a side-panel conversation becomes inefficient.
- From conversation to action: The user is no longer just asking questions. They are trying to complete work.
- Embedded AI vs chatbot detours: Instead of asking users to leave the workflow and translate everything into prompts, intelligence should increasingly appear inside the work itself.
- The new design challenge: Designers must now shape intent, previews, visibility, reversibility, autonomy and human-AI collaboration.
A practical framework for product teams
Ask these five questions in every AI workflow:
1. Is chat truly the best interface here, or just the fastest one to imagine?
2. What is the actual user job to be done?
3. What context already exists in the product that AI can use without asking users to repeat it?
4. What can the system safely do on its own, and what needs preview or approval?
5. How will the user understand, verify and recover from the system’s actions?
If more teams asked these questions, AI products would immediately become better.
Chapters
00:00 Why the chatbox may no longer be the default
01:08 Why chat won in the first place
02:35 Why chat still works for some AI use cases
03:09 When chat starts to break
04:10 From conversation as interface to action as experience
05:04 Why chat relocates friction into language06:03 Why this is a huge design opportunity
07:03 Intent instead of commands
08:09 Why the interface is not disappearing
09:02 What the future of AI interfaces probably looks like
10:09 When chat is still the right pattern
11:26 The UX patterns agentic systems need
11:39 Intent previews
12:30 Rationale and understandable action
13:05 Status and progress communication
13:33 Audit, undo and reversibility
14:31 Adjustable autonomy
15:26 Chatbot sidebar vs embedded AI workflow
16:43 What designers need to learn next
17:47 Five questions product teams should ask
18:03 Why this shift should make designers curious, not anxious
19:23 The future of AI interfaces: embedded, contextual, trustworthy
--
Links
Website: https://humanxintelligent.com
Join the conversation: https://forms.gle/HLAczyaxqRwoe6Fs6Li...
LinkedIn: / madalenafigueirasdacosta
Instagram: / designwithmaddie
--
🎙️ Human × Intelligent explores how humans and intelligent systems evolve together across AI product design, agentic systems, UX strategy and the future of work.
Subscribe for more on:
AI interfaces
agentic UX
AI product design
human-AI collaboration
embedded intelligence
future of AI workflows
---
#AIUX #AIProductDesign #AgenticAI #HumanAI #HumanXIntelligent