Inspiration

The Startup came together after the team ran in a hackathon and won a position in the innovation section of Pontifical Catholic University of ParanĂ¡. From then on, the group has created solutions to help everyone live in a connected and smart way. The globe is currently facing a pandemic of enormous proportions that will change how people live, communicate, and interact with each other. In this context, Agatha, meaning automated gastronomic assistant totally helpful and accessible, was born. Through her, the group aims to not only make ordering at restaurants quicker, easier, and more fun, but also safer because of the reduced human contact, essential in the current situation.

What it does

Agatha will be available in the tables of the establishment either through QR codes or tablets. Through these means, the clients will be able to access the menu and self-order everything they need with Agatha. Her uniqueness comes from the fact that she will have a personality, this makes ordering not only easier, but also more interesting and fun. In addition, she will be able to recognize and communicate through sign language, increasing the restaurant's accessibility to nearly everyone. With this tech, the interaction with the waiters will reduce considerably, minimizing the probability of getting into contact with the Coronavirus, and reducing expenses for the establishment.

How we built it

Agatha was programmed in js, react.js, wit.ai and css. We began by programming the Speech to Text and Text to Speech part, obtaining the functionality of the computer being able to understand what is spoken and accomplish a specific task. with this in mind, we noticed that Wit.ai would be an extremely useful application. Through Wit.ai, we were able to create not only a vocabulary for Agatha, but also a whole "context" for her to act in, akin to a real waiter. With this, Wit.ai would recognize the intention of the user's input(be it spoken or written) and send it to our application. Within our application's code, there is a function containing a list of possible answers (chosen randomly) for each registed intent. With this, Agatha will answer the user appropriately. In order to view the menu, the tab of each client, Agatha's image, the chat and the other features of the application, a program in css and react.js was created.

Challenges we ran into

During our project's development, we had difficulties with the language, since no one within our group had extensive experience with it. Structing the connection between Wit.ai and our application was challenging, since we had to develop two AIs, connecting them in such a way that one receives the client's intentions and commands, while the other analyses and separates each order correctly and systematically.

Accomplishments that we're proud of

it took us an amount of effort and racked brains to develop a program with ample functionalities which, after a certain amount of training, will be able to become something helpful and useful to a signficant group of places. Through the ample support and training we had, we could develop, in a way, an AI which is already recognizing the user's will.

What we learned

To get into the market we had to research about how important the food industry is to the world. After ample research and discussions, we understood that even this industry, which is billionaire, needs innovation. With this in mind, we started to work in an AI that would revolutionize the market. to accomplish this, we learned how to utilize tools such as react.js and css to build a website. Nowadays, just a website is not enough to get by on the environment of the connected world, so we also had to study other important tools, especially Wit.ai. With this tool, we understood how to connect it to our code, and how to work with two simultaneously.

What's next for Agatha

In the near future, the group plans to transform Agatha into a hologram. With this tech, ordering will be very unique, getting the attention of more customers, making their experience in the restaurant special.

Built With

Share this project:

Updates