AI in the design workflow: tools we use (and tools we skip)

Every design conference now has an AI panel. Every client asks about it. The honest answer is that AI is a genuine accelerator for some tasks and an expensive distraction for others.

At Konva we use it selectively — always under senior oversight, never as a stand-in for judgment.

Where AI helps

Research synthesis: clustering interview transcripts and survey responses far faster than coding them by hand.

Content exploration: spinning up headline variants and microcopy drafts for a designer to sharpen.

Asset production: background removal, upscaling, and quick moodboard generation that would otherwise eat an afternoon.

Code scaffolding: turning design tokens into starter component code an engineer can build on.

Where we skip it

Final UI design: generated interfaces lack the contextual refinement that only comes from understanding the user and the brand in front of you.

Strategy and positioning: these lean on human empathy, commercial instinct, and the messy work of bringing stakeholders along.

Client presentations: authenticity matters here. We present our thinking, not a machine's approximation of it.

Our policy

We publish our AI policy openly and talk through its use on every engagement. Client data never enters a public model without explicit consent.

The goal was never to use AI everywhere. It is to use it in the handful of places where it measurably improves the outcome for the client.