The Family Meal Planner is an example of Agentic AI and vibe coding to solve a problem. Each week we went through the same routine of viewing our pantry for what was available, going to the internet and finding recipes for a certain set of meals, shopping for what we were missing. Repetition, preferences and constraints, inefficient manual current process. Ripe for change.
Use case and inputs: 2 adults, pescatarian and vegetarian meals. Cuisines we enjoy: Latin, Asian, Mediterranean. Don't mind leftovers.
Task break down: Intake constraints and preferences. Analyze current pantry. Generate Recipes for approval. Generate Shopping list based on approved recipes. Update pantry as needed

Using Claude as both the agent and method for coding, I created a backend on google sheets. Loaded 2 sets of goods, a live look at pantry items to be used for these meals and a list of staples it could consider always on hand. I then created the agent prompts to intake “what it does” and our constraints and preferences, I continued to iterate to allow state sharing with my wife, so she could use it as a web app on her phone. The result: A meal app that uses real time data and personal preferences to create weekly meals, generate a grocery list with accurate location based prices (Kroger API) and takes in feedback on recipe quality and uses prompts to auto-update the pantry based on what you bought and what you made.