Why Dishire Was Designed

1-1. Recipe platforms naturally evolve toward AI-based systems

Recipes have continuously evolved with media platforms — from books and TV to blogs, YouTube, and short-form content. I viewed generative AI as a natural next step in this evolution, and expected recipe services to eventually migrate toward AI-driven platforms.

The initial idea behind Dishire was simple: a service where users could input their situation, constraints, or available ingredients, and automatically receive a suitable meal recommendation.

1-2. The real difficulty lies in decision-making, not recipes

Through this project, I realized that what people struggle with most is not the recipe itself, but deciding what to eat. Traditional recipe platforms assume users already know what they want, leaving the decision stage unsupported.

Dishire reframes this problem as a Food Decision Maker, helping users choose before they cook.

1-3. Constraints make food decisions even harder

Users with dietary constraints — allergies, vegan diets, low-carb plans, or weight management goals — face more complex decisions.

One of Dishire’s core goals was to help users find satisfying meals while respecting such constraints.

2. Generative AI excels at handling context and constraints

Large language models like GPT are not limited to retrieving existing recipes. They can flexibly generate new meals based on textual descriptions of context, preferences, and constraints.

• Contextual inputs such as mood, time, and available ingredients
• Simultaneous consideration of difficulty and dietary goals
• Generation of situational meal suggestions rather than fixed recipes

In Dishire, user context and profile information were embedded into structured prompt templates, allowing the LLM to generate recipes aligned with those conditions.