PurePicks AI supplement advisor (concept)
A self-initiated concept for an AI that recommends supplements and is open about health risks and about what it doesn't know.
5
trust patterns designed
TL;DR
- Question
- How might an AI recommend supplements and stay honest about its limits, especially for people who take medications or have a health condition?
- What I did
- Designed how the AI explains its picks, admits uncertainty, stops for safety and shows ingredients.
- Status
- Concept and interactive prototype.
Why I built this
The supplement market is full of marketing claims and confusing labels. AI can help people sort through it, but a confident wrong answer about health can hurt someone. I wanted to design how an AI explains itself, admits when it isn't sure and tells you to see a professional.
Trust patterns
- Explain every recommendation. Each suggestion shows the goal it matches, the key ingredient and how strong the evidence is (strong, mixed or limited).
- Show uncertainty. When the evidence is weak or the user's answers are incomplete, the AI says so.
- Stop when it should. If a user mentions medications, pregnancy or a condition, the flow flags possible interactions and recommends talking to a professional before continuing.
- Ingredient transparency. A plain-language breakdown of what's in each product, with doses compared to common reference values and third-party testing status.
- User control over data. Users choose what health information to share and can see how it changes the results.
