Move-In Assistant

summary.

Move-In Assistant uses AI to read senior living admission forms and fill in the resident's profile, with staff checking every field before it's saved. It replaces hours of re-typing across roughly 45 documents per move-in.

Sole product designer · 1 product lead, 1 AI engineer, 3 developers · Concept to pre-launch over two rounds of customer testing

problem.

Staff re-typed resident details from admission forms by hand. It was slow, error-prone, and there was no easy way to check new information against the existing profile.

what I did.

• Turned silent auto-fill into review-first. Customers kept saying "I need to see what the AI is doing," so every extracted field is checked before it touches a resident's record.
• Won confidence scores per field, at a higher bar. I pushed for a score on every field instead of one per page, and raised "high confidence" from 85% to 95% for clinical data.
• Got Legal's full-page AI warning down to one sentence by working with our AI engineer on a proposal to train the model without personal health information.
• Designed for when the AI fails: a clear state for every step, undo, and an audit trail built on existing reports instead of new infrastructure.

impact.

83% usability success across the core workflows, and 11 customers confirmed they'd adopt it. The AI model, back end and front end are built.