Snap, Chat, Cook: Redefining Meal Prep with Visual AI and Conversational Agents
Role: Product Designer & Creator Core Tech: Computer Vision, LLM (Conversational AI)
The Problem: "The Fridge Stare"
We have all been there. You open the refrigerator, stare at a random assortment of ingredients—half a cabbage, some eggs, a block of tofu, and leftover chicken—and ask yourself: "What can I actually make with this?"
Existing meal planner apps usually fail at this specific moment because they require manual entry. You have to type in every ingredient, which is tedious friction that leads users to abandon the app and order takeout instead.
The Solution: Frictionless Intelligence
I designed an AI Meal Planner that removes the manual data entry and replaces it with visual context. The core loop is simple: See, Configure, Converse.
Instead of searching for recipes, the user provides the context (ingredients), and the AI Agent builds the bridge to the final meal.
Key Features
1. Visual Discovery (From Camera to Context)
The entry point of the app is the camera. Users simply snap a photo of their open fridge or the groceries on their counter.
How it works: The app utilizes image recognition to identify ingredients instantly.
The UX Benefit: It eliminates the cognitive load of inventory management. The app knows what you have before you even list it.
2. The "Smart Setup"
Before generating a plan, the app asks for high-level structural preferences to frame the output:
Duration: "Plan for the next 3 days."
Frequency: "I need 2 meals a day (Lunch and Dinner)."
This creates a skeleton schedule that the AI will fill with culinary content.
3. The Core Differentiator: The AI Agent Chat Room
Most meal apps give you a static list of recipes that you can't change. I took a different approach. Once the initial plan is generated, a dedicated Chat Room is created with an AI Chef Agent "inside" the plan.
This turns meal planning from a one-way notification into a dynamic conversation.
The Agent Capabilities:
Cuisine Styling: The user can say, "Make these ingredients taste like Indonesian food," or "Switch Tuesday's dinner to Italian style," and the Agent instantly rewrites the recipes and spice lists.
Nutritional Negotiation: Users can set specific guardrails via chat, such as "Keep every meal under 500 calories," or "I need high protein for Wednesday post-workout."
Recipe Adjustment: If the user realizes they don't have an oven, they can simply tell the agent, "I don't have an oven, change the cooking method to stove-top," and the plan updates in real-time.
Why This Design Works
By combining Computer Vision with Conversational UI, we solve two massive user problems:
Input Friction: The camera solves the "I don't want to type" problem.
Rigidity: The Chat Agent solves the "I don't like this specific recipe" problem.
This app doesn't just tell users what to eat; it collaborates with them to create a plan that fits their pantry, their palate, and their nutritional goals.