The shift is not about adding AI. It's about rethinking interaction
Every major technology shift has demanded that designers reconsider their fundamental assumptions. Mobile didn't just make screens smaller. It changed when, where, and how people interacted with software. AI is doing something similar, but the implications are deeper.
AI doesn't just add capabilities to existing interfaces. It changes the relationship between user intent and system response. As designers, we need to develop new mental models for this.
Three principles I've found useful
1. Design for legibility, not magic
The temptation with AI is to make things feel effortless, hiding the complexity behind a seamless experience. But in enterprise contexts, users need to understand what the system is doing and why.
I've found that the best AI-powered interfaces make their reasoning visible. Not through lengthy explanations, but through clear signals that help users build accurate mental models of the system's capabilities and limitations.
2. Build trust through controllability
Trust in AI systems isn't built through accuracy alone. It's built through control. When users can adjust, override, and refine AI behaviour, they develop confidence in the system even when it makes mistakes.
In my work on admin experiences at Atlassian, this meant designing AI recommendations with clear provenance, easy dismissal, and explicit feedback loops. The AI suggests; the human decides.
3. Start with the decision, not the data
AI can surface vast amounts of information. But information without context is noise. The designer's job is to start from the decision the user needs to make and work backwards to determine what AI-generated insight would actually help.
This is the opposite of the "let's show everything the model knows" approach. It requires deep understanding of user workflows and the specific moments where AI can reduce uncertainty or cognitive load.
What this means in practice
These principles aren't abstract. They manifest in concrete design decisions: the placement of confidence indicators, the design of override mechanisms, the way AI-generated content is visually distinguished from human-authored content.
The designers who will thrive in this era are those who can hold two things simultaneously: genuine excitement about AI's capabilities and rigorous attention to the human experience of interacting with intelligent systems.
The future of enterprise design isn't AI or human. It's AI making humans more capable, with designers ensuring that capability is accessible, trustworthy, and genuinely useful.