Maple Scribe

Voice-powered healthcare documentation

Company: Maple
Role: Product Design Lead
Timeline: Six weeks, 2025

Research
Voice UX
Component system
Interface design

The project

Maple Scribe completes the Maple suite: AI voice documentation for healthcare providers. As Product Design Lead, I designed a new approach to care records, where caregivers capture what happened by talking to Mei, an AI assistant, in plain language. I explored and prototyped the voice AI concepts, the entry type system and the full UI/UX framework before the project transitioned.

The problem

Healthcare documentation has to be fast, accurate and compliant, all at once, and it weighs on caregivers. I ran field studies with caregivers in assisted living, home care and hospitals.

of a caregiver’s time can go to paperwork: 40%
core kinds of record, each with its own data: 11

01Paperwork takes up to 40% of their time.See how it was solved
02Each of the 11 kinds of record has its own data.See how it was solved
03Speaking is not always an option: the setting or privacy can rule it out.See how it was solved
04Records have to be accurate and compliant, with no errors or missing information.See how it was solved

How I solved it

Voice documentation that takes paperwork off caregivers while keeping records complete and accurate. The screens are from the prototype; its caregiver and residents are sample data.

Log care by talking to Mei

Caregivers log care hands-free by talking to Mei, an AI assistant, from vital signs to incidents. Conversational flows understand medical terms, context and common documentation patterns, so one spoken sentence can become several entries.

The latest entries for each resident, with Mei one tap away to log new ones.

One sentence, three entries

Entries Mei created from one spoken sentence: temperature, bathing and Ibuprofen, under the caregiver’s words

One component per kind of entry

I mapped the core documentation types and when caregivers can realistically log them. Each type gets a modular entry component: numeric inputs for vitals, selections for assistance levels, text for descriptions and dedicated inputs for medications. Every flow ends the same way: open the entry, edit it, or delete it after a confirmation.

Bathing: one picker, the assistance level, from independent to refused help.

Bathing entry flow: assistance level picker, edited entry and delete confirmation

Scroll sideways to follow the flow

Voice, touch or both

Caregivers can use voice, touch or both, depending on the setting and on privacy. Each entry component works by voice and by hand, and any value can be corrected on a focused keypad.

Temperature entry with the route left as Select

Nothing was said about the route, so the temperature entry waits on a dashed Select instead of a guess.

Checked before it goes on the record

Mei flags missing information and asks the caregiver to review every entry before it is saved, and nothing is deleted without a confirmation. Scribe was designed for HIPAA and medical record accuracy standards.

Review of the entries Mei created: temperature, bathing and an expanded Ibuprofen medication entry

Outcomes & learnings

The project transitioned before full implementation, but its design framework set the standard for voice documentation across the suite.

Projections, not measured

Voice entry was estimated to cut documentation time by 70% compared with traditional charting, potentially saving caregivers more than two hours per shift. Structured entries with validation rules were designed to raise documentation completeness by 85% and to cut errors with prompts and verification steps.

Scalable architecture

New entry types and care settings can be added without redesigning the core.

Influence

Voice-first principles from Scribe were adopted across the other Maple products, Maple Care’s AI assistant among them.

Maple Scribe: Ask Mei
Maple Care: Ask AI

Book a call Back to all work

Next case

Eletivo

Four products, one designer, twelve weeks