This is a submission for the Twilio Challenge
What I Built
Ever had a craving for a particular meal, but had no idea where to find it or who sells it at a reasonable price?
Where I'm from, it's easy to find meals such as pizza, or maybe a burger and french fries, but harder to find which restaurant serves a great T-Bone steak.
To help to solve this issue, I created BellyFull, a simple AI powered chatbot that recommends meals from restaurants around the island of Nassau, Bahamas based off a user's input, using the WhatsApp Sandbox mode in Twilio Programmable Messaging along with the Gemini AI API.
Technologies/Frameworks used
Twilio Programmable Messaging along with a Whatsapp Sandbox serves as the communication medium for the end user and chatbot.
Gemini is used to parse through restaurant menus to find meals, as well as maintain the state of the user/chatbot conversation.
Express.js serves the Twilio webhook that's triggered when a user sends a message to the chatbot.
Supabase is used to store restaurant information, menus, location and addresses.
Vercel is used to host the chatbot
Demo
Here's a video of how it works
How to try it out
You can either whatsapp the number +14155238886 with the message join-something-season or scan the QR code below to get started.
Once you're connected, start typing to start the conversation.
When you're finished, type "stop" to disconnect from the sandbox.
Twilio and AI
Twilio makes it very easy to create interactive experiences through various messaging platforms. With the help of Gemini AI, I was able to create a product that can create dynamic responses on the fly, based off of a user input. Instead of having to craft specific conditional statements to answer end users, I was able to structure a prompt and let generative AI handle the rest!
Additional Prize Categories
I don't feel as if this submission meets any of the additional categories. I'd nonetheless be honoured if nominated for any of the other prizes.
Top comments (0)