Case study
Can This Hurt Me?
Designing a mushroom assessment app around uncertainty
Independent project · 2025–2026
Role
Product design, UX, AI-assisted development, product strategy, creative direction
Timeline
2025–2026
Scope
Product strategy, interaction design, AI output governance, voice and copy, App Store launch
Delivery
App Store MVP
Designing around uncertainty
A plausible answer can be misleading when someone wants to know whether a mushroom is safe. I built Can This Hurt Me? around that tension: helping people understand possible risk while withholding permission to eat a mushroom based on an AI assessment. I owned the concept, interaction design, language, visual system, output rules, and launch, using AI-assisted development to build the product.
Early research mapped mushroom hunting as both culture and public-health risk: part field guide, part folk practice, part romance, part emergency warning.
Keeping assessment separate from permission
The product treats the model’s response as an input to the experience. Before displaying a result, the app considers the available photos, uncertainty, and potential risk signals. The interface distinguishes a possible match from eating guidance. Its states include a limited assessment, caution, a clear Do Not Eat warning, and access to emergency help. The intent is to prevent a convincing identification from being interpreted as proof of safety.
Asking for better inputs
The capture flow requires multiple photo angles before assessment. That requirement makes the limits of a single image visible within the interaction itself. The result hierarchy brings the risk summary, uncertainty, eating guidance, and lookalikes forward. Supporting detail remains available without competing with the primary message.
Making caution readable
The design uses different levels of caution rather than giving every result the same warning. Language, hierarchy, and visual treatment work together so color does not carry the meaning alone. Large targets and a clear information order support field use. The question for every screen was whether someone could misunderstand the result as reassurance.
Product states
Selected screens from the App Store MVP.
The MVP flow distinguishes photo capture, assessment, risk guidance, emergency help, and support. A possible match is never permission to eat a mushroom.
Delivery and next evaluation
The MVP is available on the App Store. Subsequent work strengthened the rules so potentially lethal candidates trigger a Do Not Eat result rather than a quieter uncertainty state. These are product safeguards, not evidence that the app can establish whether a mushroom is safe to eat. Further evaluation of interpretation, failure cases, expert feedback, and model robustness remains important to the product’s development.