Inspiration

Appointment handling is a time-consuming process that most smaller private businesses, such as doctor offices, have to deal with, and we saw the potential for automation in this pipeline. Our tool provides both monetary and productivity benefits to the business owners and their secretaries, allowing them to focus on the tasks that matter most.

What it does

Noemi is a virtual assistant that recognises speech and aims to accelerate and simplify the appointment booking process, while also being available 24/7.

How we built it

We used Python and the pre-trained whisper model (by OpenAI) to translate speech into text. Then, after interacting with the user to decide on the best available time slot, our assistant creates an event in the business owner's calendar if feasible.

Challenges we ran into

The requested slot may not be available so the system needs to interact with the user and suggest a feasible time slot.

Accomplishments that we're proud of

We worked with the required libraries for the first time and we managed to build a working MVP in a short amount of time.

What we learned

We became familiar with state-of-the-art speech-to-text and text-to-speech models and through our search identified that the problem is far from being solved in practice, so there is a potentially large market for our idea.

What's next for Noemi

We plan to integrate a more sophisticated NLP model that can understand the complex nature of the temporal attributes of the users' requests, and integrate our model with a hotline-as-a-service app that can scale to multiple requests in parallel.

Built With

Share this project:

Updates