Implementing AI into an existing app

For this project, I designed a conceptual feature to enhance the Cookpad app by allowing users to discover recipes through photos. This feature aims to help users identify and recreate dishes they encounter in restaurants or social settings, making recipe discovery faster and more intuitive.

Many users come across dishes they love but don’t know the names or ingredients, making it hard to search for recipes. Traditional recipe discovery often requires detailed search terms, which may not be possible if the user is unfamiliar with the dish. My goal was to streamline this process by designing a feature that enables recipe discovery via dish photos.

• User Engagement: Based on prototype testing, users appreciated the idea of quick recipe discovery through visual input, especially with the option to refine AI suggestions.

• Enhanced Exploration: The concept encourages users to explore and recreate dishes they might not have initially planned to, broadening their cooking repertoire.

• User Control: The option to manually select from multiple suggestions mitigated potential frustrations with AI misclassification.

• Designing for AI Limitations: Although the feature concept relies on AI, I focused on building a user-friendly experience that allows flexibility, even if AI recognition isn’t perfect.

• Balancing Simplicity and Guidance: Some users needed context to understand the selection process. Adding brief tooltips and a mini-tutorial improved clarity and onboarding.

The proposed feature allows users to take a photo of a dish, which the app would analyze to suggest potential matches. If the AI isn’t entirely certain, the feature presents up to four possible dish options (e.g., taco, taquitos, burrito, enchilada), allowing users to select the one that best matches their dish. Once chosen, the app displays a recipe with photos, ingredients, and step-by-step instructions.