Claude Cancels Cavities

Inspiration

We chewed on a big question: how can we help people brush up on their mindfulness and keep up with what’s going on in the world without adding more to their plate? Everyone’s brushing their teeth (we hope!), so we thought. What if your toothbrush could speak volumes? With Word of Mouth, we're turning your daily rinse-and-repeat into a moment that actually sticks with habit stacking.

What it does

Let’s face it—there just aren’t enough hours in the day to do everything: stay informed, practice gratitude, check in with friends, and maybe even breathe for a second. That’s why Word of Mouth is here to fill the gaps. Every morning and night, your toothbrush gives you a personalized two-minute audio experience while you brush. Morning routines include a tooth-sized news brief, stock updates, weather, and calendar highlights to get your day off on the right bite. At night, it scrapes away the plaque of the day with guided reflections, gratitude prompts, and even fresh updates from your Brush Network—a place where you and your friends can share weekly updates to hear while brushing.

How we built it

Backend:

Set up the APIs for news, stocks as well as setting the reflective questions
After receiving relevant information, we personalized the information summary conveyed to the user based on the intake experience
Set up voice-to-text with ElevenLabs to convey the summaries with different types of voice tones
Set up a Firebase database to collect user profiles Frontend/User Interface:
We built an interface that takes in demographic and qualitative data for the user
We constructed a toothbrush with a speaker attached

Challenges we ran into

We initially kept getting a CORS error when trying to connect the stock API
Web scraping from news on pay-walled webpages
The news API pulled news that wasn’t current
Firebase database issues integrating with python and javascript functions — had to move everything onto a flask frontend and backend
Integrating the different pieces of functionality was a challenge.
Prompt engineering different tones, personalities, and responses without sounding like a conversation was difficult

Accomplishments that we're proud of

Have a working, talking toothbrush (Physical proof-of-concept)
The toothbrush pulls from different APIs to give the user curated information based on their preferences

What we learned

Better understanding of Claude’s features
Better understanding of how to build a data pipeline
How to work with voice synthesis tools like Eleven Labs and personalize tone to make it sound human, friendly, and bite-sized.