A system for families managing sensory-sensitive eating
Sensory Sprout is a meal planning system designed for families where dinner does not reliably work.
Most tools help you choose recipes. This system models whether a meal will actually be eaten based on how it is prepared, served, and experienced by the people at the table.
To do that, it captures each family member’s constraints at a level typical tools ignore. Not just ingredients, but texture, temperature, preparation, and predictability. These signals are then used to select and adapt meals that can work across the household.
I designed and built the full system end to end: user research, conversation design, prompt architecture, evaluation framework, product design, ranking logic, transformation logic, and trust architecture. Self-initiated, solo.
I have to make 3 different meals every night… and sometimes they won't eat more than two bites.
Anonymous · Local mom's Facebook groupMemory for recall, not reliving
People want to look back and remember important things, rather than re-experiencing them.
Surface relevant suggestions
Users want personalized prompts, reminders, and tasks — both in the moment and as a recap at the end of the day.
Targeting students
Massive AI adopters, with a familiar form factor.
Less competitive space
Fewer constraints on battery life and data plans.
On-the-go form factors
Fit for mass adoption.
Discover
Map the problem space.
Define
Frame the right question.
Design
Explore solutions.
Evaluate
Test and measure what works.
Caption describing what this image shows.
Caption describing what this image shows.
Caption one.
Caption two.
Caption three.
Turn inputs into a decision model
Constraints, patterns, and execution rules are structured for decision-making — not storage.
The system can treat hard rules differently from softer tendencies.
Choose meals that can work
Meals are ranked by viability, not appeal: structure, branchability, adaptability, and familiarity matter more than ingredient match alone.
Turn inputs into a decision model
Constraints, patterns, and execution rules are structured for decision-making — not storage.
Choose meals that can work
Meals are ranked by viability, not appeal: structure, branchability, and familiarity matter most.
Rank with confidence, not certainty
Strong signals can guide decisions without overcommitting to weak or early inferences.
6
Parent interviews conducted
4
Adaptive flows for different eater types
6
Parent interviews conducted
4
Adaptive conversation flows designed
15/2/0
Pass / Partial / Fail on strongest config
Optional caption. Use this to explain what the image shows and why it matters to the narrative.
| Layer | Failure | Cost | Design |
|---|---|---|---|
| Retrieval | Allergen shown | Irreversible | Deterministic |
| Ranking + Transformation | Bad execution | Recoverable | AI + rules |
| Assessment | Wrong constraints | Adaptive | AI + routing |
| Configuration | Pass | Partial | Fail |
|---|---|---|---|
| 4o-mini · old prompt | 6 | 8 | 3 |
| GPT-5.2 · old prompt | 9 | 7 | 1 |
| GPT-5.2 · restructured prompt | 15 | 2 | 0 |
Parent completes intake
The system asks about each family member's tolerances, textures, and restrictions through a guided conversation.
System builds a preference model
Constraints and patterns are structured into a decision model — not just stored as preferences.
Meals are ranked by viability
The system surfaces meals that can actually be prepared and adapted — not just ones that sound good.
Intake conversation
Guided questions build a profile of tolerances and restrictions.
Decision model
Constraints are structured for ranking and adaptation.
Viable meal plan
Meals ranked by viability, with adaptation paths for each person.
Discover what actually fails at dinner
Interviews with families revealed the failure happens at serving, not planning — the wrong frame for most tools.
Reframe from preference to tolerance
The real constraint isn't what people like — it's what they can tolerate across texture, predictability, and preparation.
Design for the last mile, not the first choice
The system succeeds when dinner actually gets eaten — not when a recipe gets saved.