Inspiration
The race of big tech companies to build the first Individualized convolutional neural network! and the current personal assistant AI Agents being less capable than expected.
What it does
AI Assistant that schedules your calender events and todos all through natural language processing.
How we built it
I built the entire backend system with no code using make.com ai automations allowing me to route webhooks from our AI Voice Agent for specific tasks ( bookcalender, checkavailability, createTodo ) i built these functions within make.com running in the background 24/7 and triggered when our AI voice agent triggers the webhook for each, prompt engineered the AI model to undestand when to schedule todos, create caledner events and how to call the webhooks. we then schedule all the calender events on google calender and todos on a google sheet and then use the google API to take the calender data and put it in our frontend in realtime, alongside the google sheet data and display that in our frontend in real time.
Challenges we ran into
google API with the OAuth token was a very long inconvenience
Accomplishments that we're proud of
What we learned
how to pull data from the google API, automation using AI and no code to completely automate and tie together a backend system. and how to build a real publishable web application
What's next for Day2Day
With just the Todos and Calender events being completely managed through Natural Language, we can polish this system and add a database to save every conversation as memory, further tuning the model off every conversation allowing it to adapt and grow to be your own personalized assistant overtime. with these 3 features( continous learning, todos, calender) our AI Agent functions similarly to your own personal convolutional neural network with much less complexity, we can setup accounts and billing and turn this into a real, practical and real world applicable SAAS application.
Log in or sign up for Devpost to join the conversation.