Innovating the future of AI
in pregnancy nutrition
0 to 1 Speculative Design
AI + Wearable
16 weeks
Product Mira.ai
A nutrition and nausea-management system for pregnant mothers that pairs an Apple Watch wearable with an AI-powered companion app. Mira senses physiological signals, adapts meals to how she's actually feeling, and offers on-demand relief — meeting women in the moments between doctor visits, where most real decisions happen.
Details
Mobile app + Apple Watch
surface
Product Designer
role - passion project
Pearl S (designer) + Hugh.D (mentor)
team members
AI Workflows ∙ Visual & Motion Design ∙ Desk Research ∙ User Testing
scope
Problem The Body Gets Forgotten
Most pregnancy apps count kicks and bump sizes — but miss the realities of nausea, discomfort, and shifting nutrient needs. Women get 10-minute doctor visits and 1 minute of nutrition guidance, then navigate 9 months of real decisions alone.
Solution A Companion who can Sense Needs
A wearable-app system that combines clinical insight, predictive sensing, and a custom Apple Watch–compatible strap — sensing nausea before it intensifies, adapting meals to symptoms in real time, and keeping every recommendation human-approved.
Result Projected Impact
A human-in-the-loop concept designed to ease nausea earlier, close nutrient gaps, and rebuild trust in digital pregnancy guidance — replacing fragmented Googling or vague advice with one coherent companion.
Projected Impact validated in user testing
15/15
Testers reported greater confidence
Every tester felt more informed and in control during pregnancy decisions
100%
Preferred prescription uploads + context-aware AI
Over traditional pregnancy-tracking solutions
~70%
of pregnancies affected by nausea
Global meta-analytic rate — 69.4%, ranging 35–91% (Einarson et al., PubMed, 2013)
CONTEXT
Mira.ai is a nutrition and nausea-management system for pregnant mothers.
It started because I kept seeing the same pattern around me: friends, cousins, classmates getting pregnant and the support system around them was failing them. Their doctor visits were 10 minutes. Nutrition guidance was one minute. Most of the real decisions were happening between Google and Reddit threads, alone at 2 AM.
PROBLEM
Existing solutions track the baby but ignore the body carrying it.
Scientists & engineers were stuck juggling too many tools, delaying cancer patient care. This is what they were dealing with,
Waiting a long time for systems to process
Dealing with the cognitive strain due to tool-hopping
GOALS + NORTH STAR
To design a companion + digital twin that senses what the mother is feeling, explains what it knows, and acts only with their consent, so pregnancy feels less uncertain and less lonely.
Sense Early: Use HRV and wearable signals to detect nausea before it escalates.
Relieve Safely: Offer median-nerve stimulation on demand — clinically precedented, always human-approved.
Personalize Deeply: Build a bio-digital twin from her own prescriptions, blood work, trimester data, and dietary restrictions — not a generic model.
Earn Trust: Explain every recommendation in plain language, and keep her in the loop for every intervention.
PROCESS + KEY INSIGHTS
I co-owned product strategy and design with fellow designer. Over 16 weeks we conducted desk research, user interviews, and testing, mapping the gap between what medicine offers and what women actually use day-to-day.
Key Insight 1: Nausea is predictable Studies show that commonly tracked metrics like heart rate and HRV fluctuate before nausea hits. That gave us a sensing window no existing pregnancy product uses.
Key Insight 2: Nausea can be eased Stimulating the wrist's median nerve with electrical pulses is an FDA-precedented way to reduce nausea — but existing devices require manual activation after symptoms start.
Key Insight 3: Introducing a smart strap We designed a custom Miraai strap that integrates directly with the Apple Watch, embedding the median-nerve activator into a form women already wear. That turned a clinical device into an everyday companion — and opened the harder design question: should it act without her permission?
SOLUTION
Mira.ai, a nutrition and nausea-management system that adapts to a mother's changing needs.
TECH FRAMEWORK
By analyzing symptom logs and wearable signals, Mira.ai builds a bio-digital twin of the mother — identifying nutrient gaps and symptom risks, and offering real-time guidance through the app and smartwatch.
The system in three layers:
Inputs — Mobile app (manual symptom logging, medical data) + Smart Watch (wearable sensing)
Intelligence — Data training (bio-digital twin), predictive analysis (nutrient gap + symptom risk), conversational AI agent
Outputs — Mobile app (meal & supplement suggestions, adaptations) + Smart Watch (updated nausea risk)
This is where I started thinking like both a designer and a system architect: What data does the model need? What should be retrieved from her medical record? What must not leave the device? Meal and supplement pairings would come from a structured, safe dataset and not a hallucinating model.
DESIGN 1/6
On the Wrist
Three Screens, Three Moments
The watch does the quiet work. The phone is for planning; the wrist is where Sneha is met in real time.
1. Detect — a passive notification slides in alongside her other pings. Mira whispers, doesn't alarm.
2. Decide — full-screen takeover, two actions: Activate or Dismiss. No snooze, no third path.
3. Deliver — the pulse engages. The Miraai mark appears only now — because this is the only moment the product is acting on her body. One stop button, centered, always in reach.
DESIGN 2/6
Onboarding
Uploading the Medical Self
Before Mira can personalize anything, the mother uploads her prescriptions, blood work, trimester data, and dietary restrictions. This is the foundation of the bio-digital twin and we decided that the info must never leave their device.
DESIGN 3/6
The Explaination Banner
When the system adapts the day's meals, it tells her why: "Let's eat light fibrous food today to replenish nutrients since you threw up."
If recommendations feel like a black box, users distrust them — especially during pregnancy. Transparency is the default.
DESIGN 4/6
The Adaptive Recipe List
After a nausea episode, the home screen shifts to lighter, fibrous options like steamed sweet potato, rice mushroom congee, blueberry oats, all flagged light on queasy stomach. Same structure, different content for how she actually feels today.
DESIGN 5/6
Nutrient Goals + Meal Logging
Daily macro and micronutrient goals frame the logged meals — "Boost calcium intake today to reduce muscle cramps and strengthen baby's bones." Every tracked input feeds back into the bio-digital twin.
DESIGN 6/6
The future is conversational
The mother talks and chats with Mira.ai. The agent listens to the way she describes her symptoms like "my stomach feels off, I couldn't keep breakfast down" and cross-references her history to detect what she actually needs. The interface recedes; the conversation leads.
REFLECTIONS
Designing for consent My first instinct was to automate relief activation. Killing that feature taught me that in high-stakes health products, friction can be a feature — the consent prompt isn't a UX cost, it's what makes the system trustworthy.
Thinking like a systems architect Personalizing recommendations forced me to ask questions designers rarely own: what data does the model need? what gets retrieved? what must never leave the device? This project left me a more systems-literate designer.
Speculative doesn't mean Untethered The real-time HRV-to-stimulation integration doesn't exist yet, but HRV sensing does, median-nerve stimulation does, and both are FDA-cleared. Speculative design earns its credibility when every piece is individually real.
WHAT I WOULD DO NEXT..
Close the loop with clinicians
Right now, Mira is a closed loop between Sneha and her device. The next layer is her OB/GYN. I'd design a clinician-facing view that surfaces patterns — symptom frequency, nutrition adherence, flagged concerns — so the 10-minute visit becomes a richer conversation instead of a status check.
Mira.ai's nourished mothers lived happily ever after.
fin.