Provide ingredients and Chef AI will provide recipes for you!
Inspiration
When cooking different recipes for myself, I have often found myself asking different models for recipes, and not having my hands free to type another prompt, or click the mic button. I wanted to create a platform where you could have an ongoing conversation with a model hands-free, where it could help you with any step on a recipe, or with anything that needs to be explained. Essentially, I wanted have a guide to answer questions hands-free when cooking a new recipe, as your hands are often quite busy when cooking!
What Chef AI Does
Chef AI serves as a hands-free kitchen assistant to its user, where they list their available ingredients, and it generates creative recipe ideas for them. It guides the user through their chosen recipe step-by-step using voice interaction, assisting them with any help they made need in the cooking process.
Challenges
There were a few challenges when dealing with React state management. Often, the conversation mode will switch between listening and speaking, and this would lead to refreshes in visuals displayed to the user, and lots of flickering. This took quite a few attempts, but was eventually fixed.
Accomplishments
The project created a seamless interface for voice-to-voice interaction between Chef AI and its user by integrating Elevenlabs Conversational API in order to deliver natural conversations centered around making a delicious meal and Google Gemini to generate the ingredients and instructions for that meal.
Lessons Learned
The project taught me how to take a project from Github, and have it automatically uploaded into a website on Google Cloud, through setting up a Dockerfile within Github Actions. It also taught me how to work with multiple APIs, including Elevenlabs Conversational API.
Build Details
The UI is built using React to manage real-time conversation state, and Next.JS was used to provide the web framework. Elevenlabs Conversational API was used to enable natural conversations between users and the AI, along with Google Gemini for recipe generation. The final build was deployed to Google Cloud using Google Cloud Run with a Dockerfile in Github Actions.
In the demo video, some of Chef AI's responses are trimmed in the interest of time
Built With
- docker
- elevenlabs
- gemini
- github
- google-cloud
- next-js
- react
- typescript
Log in or sign up for Devpost to join the conversation.