/
PH1 core capbility
YEARS EXPERIENCE
3 to 5 years
TYPICAL CLIENT
VP Product, Head of Design, Director of AI, CTO
NECESSARY TIMELINE
2 to 3 months
Our POV
Generative AI breaks the assumptions interface design was built on. The output is different every time, the system sounds confident when it is wrong, and the user cannot tell which they are getting. Designing for that is a different discipline from designing a form.
The features that work set expectations before the output arrives, make uncertainty visible, and give people a cheap way to recover when it is wrong. The ones that fail look impressive in a demo and get used once.
What We Do
Design the interaction, not the prompt box. What the user is trying to accomplish, where the AI enters, how much control they keep, and what the fallback is when it does not work.
Design the failure states first. Wrong answers, partial answers, refusals and slow responses are the normal case, not the edge case. We design those paths deliberately.
Build the trust signals. Where the output came from, how confident the system is, and what the person should check before relying on it.
What We'll Deliver
Interaction design for the AI feature across its real workflow
Failure state design covering wrong, partial, refused and slow responses
Trust and transparency patterns appropriate to the stakes involved
Prototypes tested with the people who would use the feature
Design specifications your engineering team can build from
Patterns documented so your team can apply them to the next feature
When This is Essential
An AI feature demos well and is not being used twice
Users cannot tell when the output is wrong, and are finding out late
Before building a generative feature into an existing product
When an AI feature is being scaled to a broader or less expert audience
When trust in the output has dropped and the model is not the cause
Combine With These Services
Rapid Prototyping & Concept Testing — test the design before it is built
AI Evals & Output Fine-tuning — measure whether the output holds up
AI Trust & Confidence Research — find where confidence breaks down
Part of Prototyping & Testing and AI Product Evaluation and Strategy & Planning.
/
Submissions