TO STEP

Prompt G:

Inspiration

We started with a simple observation: older adults and caregivers often want the same outcome, but they describe the problem very differently.

An older adult may say, “I want to keep cooking for my family,” while a caregiver may say, “I’m worried about her balance, fatigue, and how long she can stand.”

Both are talking about the same underlying goal, but traditional care planning often starts with limitations, safety concerns, or clinical tasks instead of what the older adult personally wants to preserve.

TO STEP was inspired by the idea that aging support should not only ask, “What does this person need help with?” It should also ask, “What matters enough to them that they want to keep doing it?”

Our goal was to use AI to help bridge those two perspectives and create shared goals that feel both meaningful to the older adult and useful to the caregiver.

What it does

TO STEP is an AI-powered shared care planning tool for older adults and their caregivers.

The older adult answers a short series of questions about:

what they want to keep being able to do
what has become harder
why that activity matters to them

The caregiver separately answers questions about:

what they have noticed becoming more difficult
where they currently provide assistance
what they hope the older adult can continue doing independently

TO STEP then uses AI to analyze both perspectives, identify where they overlap, and create a shared functional goal.

For example:

Older adult:
“I want to keep making Sunday dinner for my family.”

Caregiver:
“I’m worried about her balance and fatigue while standing in the kitchen.”

TO STEP:
“Maintain the strength, endurance, and confidence needed to keep preparing Sunday dinner safely and independently.”

Instead of creating a care plan around limitations alone, TO STEP helps families create a plan around what the older adult is trying to preserve.

How we built it

We built TO STEP as a web-based MVP using Replit.

The prototype is designed to demonstrate one complete interaction:

The older adult completes a short guided questionnaire.
The caregiver completes a separate questionnaire.
The responses are sent to an AI model.
The AI summarizes each person’s perspective without choosing one side over the other.
It identifies a shared goal based primarily on what matters to the older adult while incorporating the caregiver’s observations.
The users can review and edit the generated goal before accepting it.

We designed the interface around older-adult accessibility:

large text
simple navigation
one major task per screen
plain-language questions
minimal typing
clear confirmation before anything is finalized

A key design principle was that AI should act as a translator and alignment tool, not as a medical decision-maker.

Challenges we ran into

One of our biggest challenges was scope.

The aging population is an enormous problem space, and there are many directions we could have pursued: fall prevention, exercise adherence, caregiver coordination, home safety, rehabilitation, and remote monitoring.

We realized that trying to solve all of those at once would weaken the product.

Instead, we narrowed the problem to one question:

How can we help older adults and caregivers create goals together instead of talking past each other?

Another challenge was deciding what the AI should and should not do.

We did not want the model to diagnose conditions, invent medical information, or override either person’s perspective. We therefore designed the AI to:

preserve the older adult’s own words when possible
summarize each perspective separately
identify areas of agreement
acknowledge disagreements rather than deciding who is correct
allow users to edit and confirm every generated goal

Accomplishments that we're proud of

We are proud that TO STEP evolved from a broad idea about aging into a focused tool with a clear purpose.

Rather than building another generic senior wellness app, we created a prototype around a specific human problem: older adults and caregivers can see the same situation through very different lenses.

We built an end-to-end workflow that turns those separate perspectives into one shared, meaningful goal.

We are also proud that our team was able to combine perspectives from biomedical engineering, biokinesiology, and public health.

Those backgrounds helped us think about:

function and independence
caregiver burden
healthy aging
accessibility
human-centered AI
how technology can support, rather than replace, human care

What we learned

The biggest lesson we learned is that better care is not always about collecting more data.

Sometimes the missing information is simply what the person actually cares about.

A caregiver may see risk. A clinician may see impairment. An older adult may see the possibility of losing an important part of their identity.

All three perspectives matter.

We also learned that AI can be most useful when it reduces the work required to understand those perspectives rather than trying to replace the people involved.

For TO STEP, the AI is not the expert.

The older adult and caregiver are the experts on their own experiences. AI helps connect them.

What's next for TO STEP

Our next step would be to validate TO STEP with older adults, caregivers, and clinicians such as physical and occupational therapists.

We would test:

whether the generated shared goals accurately reflect both perspectives
whether older adults feel more heard and involved in planning
whether caregivers find the goals more actionable
whether clinicians find the information useful for treatment planning
whether shared, personally meaningful goals improve engagement and follow-through over time

Future versions could also include progress tracking, functional assessments, clinician dashboards, and integrations with rehabilitation or home-health workflows.

Our long-term vision is simple:

Care should not only help older adults do what they need to do. It should help them keep doing what makes life feel like their life.