AI Experience Design

What it is

Design for products where a model does part of the work. We design the interaction around the model: what the user asks, what the system shows while it works, how a result is checked, and what happens when the answer is wrong.

The hard part of an AI product is rarely the model. It is the interface where a person has to decide whether to trust it.

Who it is for

• Teams adding AI features to an existing product

• Products built around agents, assistants or generated output

• Teams designing AI workflows for internal or expert users

• Organisations with confidentiality constraints on where AI can be used

Problems it solves

The feature demos well and nobody uses it twice. Users cannot tell what the system did or why. There is no path back when the output is wrong, so people stop relying on it. The product exposes a model where it should expose a workflow.

What we deliver

• Interaction models for prompting, review and correction

• Workflow design around the model, not just the chat window

• States for uncertainty, failure and partial results

• Interface patterns for provenance, so users can see where an answer came from

• Production ready UI

• Guidance on what should not be automated

How this connects to the Lab

The studio runs its own AI experiments and publishes what they show, including a local model experiment run entirely offline on consumer hardware. The work in the Lab is where these interaction patterns get tested before they reach client products.