AI-Generated Recipe App

AI-Generated Recipe App

AI-Generated Recipe App

Client

SideChef App

Role

Product Designer (UX/UI)

Collaboration

PM, Engineering

TimeLine

Jan–May 2025

Background
People often encounter inspiring dishes when dining out — beautiful, delicious, and unforgettable.
However, recreating them at home is surprisingly difficult, especially when the dish name is unknown or the cooking method is unfamiliar.

“How can I make this?”

“How can I make this?”

“How can I make this?”

To bridge the gap between food inspiration and actual cooking, SideChef introduced two AI-powered features:
  • AI-Gen-Recipe: Generate a complete recipe from a dish photo
  • Remix AI (Substitution AI): Personalize ingredients and steps to match dietary preferences
Research & Insights
We conducted quantitative research with 412 SideChef users, combining survey data and behavioral analysis of recipe search and save events across dining-out and social media scenarios.

The results revealed consistent drop-off points driven by text-based search, while visual input, early dish recognition, and personalization significantly increased engagement and follow-through.

Across interviews, a consistent pattern emerged:

Across interviews, a consistent pattern emerged:

Insight 1

~65–70% of users start their cooking inspiration from a photo, not a recipe name

Insight 1

~65–70% of users start their cooking inspiration from a photo, not a recipe name

Insight 1

~65–70% of users start their cooking inspiration from a photo, not a recipe name

Insight 2

Over 50% of users abandon the idea of recreating a dish due to uncertainty about ingredients or steps

Insight 2

Over 50% of users abandon the idea of recreating a dish due to uncertainty about ingredients or steps

Insight 2

Over 50% of users abandon the idea of recreating a dish due to uncertainty about ingredients or steps

Insight 3

Users are 2× more likely to continue when recipes provide clear, step-by-step guidance rather than expert-level detail

Insight 3

Users are 2× more likely to continue when recipes provide clear, step-by-step guidance rather than expert-level detail

Insight 3

Users are 2× more likely to continue when recipes provide clear, step-by-step guidance rather than expert-level detail

Insight 4

Nearly 40% of users expect recipes to adapt to dietary needs such as vegetarian, gluten-free, or dairy-free options

Insight 4

Nearly 40% of users expect recipes to adapt to dietary needs such as vegetarian, gluten-free, or dairy-free options

Insight 4

Nearly 40% of users expect recipes to adapt to dietary needs such as vegetarian, gluten-free, or dairy-free options

Problem Statement

|

When users don’t know the dish name or ingredients, traditional recipe discovery fails and uncertainty causes drop-off.

Design Goals × Metrics

Goal
1
1

Start from a photo, not a dish name

Start from a photo, not a dish name

  • Photo-based entry usage +25–35%
  • Conversion from photo upload to recipe generation +15–20%

Reduce uncertainty with clear, step-by-step guidance

Reduce uncertainty with clear, step-by-step guidance

  • Progression to step view +20–30%
  • Drop-off before steps −15–25%
Goal
2
2
Goal
3
3

Make AI-generated recipes feel trustworthy and usable

Make AI-generated recipes feel trustworthy and usable

  • Save / Start Cooking actions +15–25%
  • Recipe regeneration rate −10–20%

Personalize by default with Substitution AI

Personalize by default with Substitution AI

  • Remix usage rate among AI recipes 30–40%
  • Retention uplift for dietary-preference users +10–15%
Goal
4
4
Goal
5
5

Handle edge cases gracefully

Handle edge cases gracefully

  • Non-food upload recovery rate +20–30%
  • Failed attempt drop-off −15–20%
Core Solution

AI-Gen-Recipe
（Image → Recipe）

AI-Gen-Recipe
（Image → Recipe）

After launch, 45–55% of AI-generated recipes were created through image upload or camera capture,
indicating strong adoption of photo-first recipe discovery.
Core Solution

Step-by-Step Guidance

Step-by-Step Guidance

Research showed that over half of users abandon cooking unfamiliar dishes due to unclear ingredients and steps. Clear, step-by-step guidance helps users move forward with confidence.
Core Solution

Substitution AI
（Remix AI）

Research showed that over half of users abandon cooking unfamiliar dishes due to unclear ingredients and steps.
Clear, step-by-step guidance helps users move forward with confidence.

Step 1

Connect accounts

Securely link your bank accounts and get a full, up-to-date view of your finances.

Step 2

Track easily

Nova automatically organises everything into categories without any manual work.

Step 3

Personalized tips

Get smart insights based on your habits to help you save and make confident decisions.

Results

The image-to-recipe feature drove sustained engagement
0.0mins

Avg. Time Spent per Image-based Recipe Session

Image-based generation became a core way to create recipes
0.0%

Share of AI Recipes Generated from Images

Most users progressed to full cooking instructions
0.0%

Users Reaching Step-by-Step Cooking View

Personalization was actively used, not ignored
0.0%

AI Recipes Using Remix (Substitution) AI

Image-generated recipes led to real cooking intent
0.0%

Start Cooking Rate after Image-based Generation

Impact


Results&Impact

Results&Impact

Results&Impact

Taken together, these metrics show that users not only adopted image-based recipe generation, but also spent time reviewing steps, personalizing recipes, and moving into cooking.
Taken together, these metrics show that users not only adopted image-based recipe generation, but also spent time reviewing steps, personalizing recipes, and moving into cooking.
Taken together, these metrics show that users not only adopted image-based recipe generation, but also spent time reviewing steps, personalizing recipes, and moving into cooking.

Background

Research & Insights

Problem Statement

Design Goals × Metrics

Solution & Key Features

Results & Impact

Create a free website with Framer, the website builder loved by startups, designers and agencies.