FrontDesk AI

• 

  logo

Inspiration

One of our team members had prior experience working a front desk job and recognized how AI could handle most of the responsibilities of a front desk representative. Inspired by this, we decided to create a system that automates these functions, focusing specifically on a dentistry office.

What it does

Our application, FrontDesk AI, is an AI assistant designed to perform front desk functions for a dental practice. It manages appointment scheduling for prospective patients, streamlining the booking process.

How we built it

We developed our solution using an open-source voice assistant framework called Verbi. For natural language processing, we used the Groq large language model (LLM), and Deepgram for both text-to-speech and speech-to-text capabilities. All schedule data is stored in SingleStore, and we integrated Airflow to automate database reads and writes.

Challenges we ran into

One of the biggest challenges we faced was handling AI hallucinations and generating accurate responses. We overcame this by integrating multiple Groq models, where the output from one model was fed into another to refine the final response.

Accomplishments that we're proud of

We are proud of learning about the functionalities of each of our AI tools. We are also proud of developing our overall infrastructure and having everything function as it should from a system design standpoint.

What we learned

We gained experience working with Groq, Airflow, Deepgram, and SingleStore, and learned a lot about integrating and troubleshooting connections between these technologies.

What's next for FrontDesk AI

Our next steps include integrating phone call functionality using Twilio to extend communication capabilities. Additionally, we want to expand the application beyond just appointment scheduling through Google Calendar to include features like sending personalized notes to staff via email or phone. We also aim to improve scalability so that the application can support multiple devices (10+ devices) under the same database, enabling broader deployment across practices.

Built With

• airflow
• cartesia
• deepgram
• google-calendar
• groq
• python
• singlesource