“How AI is changing what designers do – and what that means for everyone who verifies quality.

For twenty years – since screens became the primary surface of digital products – design answered one question: what should this screen look like?

AI just made that question insufficient.

The user is no longer always a person. The deliverable is no longer always a screen. And done is no longer a moment – it’s a quality bar that has to be maintained continuously, long after the Figma file is closed and the ticket is shut.

Who is actually using this? Humans, agents, voice interfaces, embedded systems – each one encounters the same design decision completely differently. Design now has to account for users it cannot interview, observe, or watch struggle.

What am I actually designing? Prompts. Guardrails. Tone of voice rubrics for AI-generated responses. Rules that generate the experience rather than being the experience. None of it lives in Figma. All of it shapes what the user encounters.

What does done even mean? When the AI output changes every session, when the model drifts silently between month one and month three, when quality isn’t a state you ship to but a bar you maintain – the old definition of done stops working. And most teams don’t have a new one yet.

This isn’t a talk about AI tools for designers. It’s a window into how design thinking is changing at the foundation – and what that means for the people whose job it is to verify that what was designed actually works.”