BabyLlama

Pitch Deck
App Screenshots
App Screenshot
App Screenshot
App Screenshots
App Screenshots
App Screenshots
Pitch Deck
Pitch Deck
Pitch Deck
Pitch Deck
Pitch Deck
Pitch Deck
Pitch Deck

Inspiration

Parenthood is one of those things that people do not get educated about much until they become parents themselves. As someone whose mother works as an educator for first-time parents, I witness this knowledge gap quite frequently. The first point of education for anyone is their family, with newborn care being extremely important for healthy development. In this way, while we were trying to build an all-purpose mobile application with AI support for baby care, we thought about many creative additional ideas that we would love to see for babies and, of course, the health of the parents themselves. For babies with working parents who cry for them, we thought about AI voice cloning for stories and lullabies... For parents who have their ups and downs with the recent addition to the family, we thought about an AI-based emotion-tracking journal with voice recording... For different scenario-based problems, we thought about a baby expert assistant... For parents who like to learn more, we thought about a trusted article provider... We also thought about many more social and necessary daily-usage features... Aiming to become the super-app for every parent for healthier development of all the babies around the world!

What it does

The mobile application has multiple features ranging from menial tasks to AI-supported creative tasks:

1. Voice Cloning of a Parent and Storytelling/Lullaby

One of the creative features of the mobile application is cloning the voice of the parent through LMNT, and voicing the stories or lullabies that are created by OpenAI. Even when the parent is away or on a business occasion, the baby can hear their voices telling stories, eventually helping them calm down.

2. Educational Article Search

Based on the user's activity within the app, we use You.com to make educative articles accessible and give them personalized recommendations on educative materials that they can read with the click of a button. This feature is especially important since first-time parents want to learn, but they do not necessarily have the time, community, and resources to look out for information.

3. Emotional Journaling for the Parents

With audio recordings evaluated by Hume.ai, we track parents' emotional conditions and their progress day by day. This is especially important after the birth to track the emotional progression of the mom, or to realize trends in the parents' emotional cycle when the baby is added to the family. We also provide an emotional dashboard for the current day.

4. Personalized Assistant

We created a specialized agent for the users that focuses on baby care. We personalized this assistant to be empathetic, understanding, and easy to understand for the parents. Using OpenAI, this agent aims to provide quick fixes, advice, and assistance in a customizable fashion when the parent has a scenario-based problem that requires contextual understanding.

5. Information Sheets, Updates, Tasks, Events and Scheduling

Though very primitive (since we did not have enough time to make them look fancy), we do small tasks such as adding tasks, and events, listing out other events, scheduling etc. We also list the information about the baby for tracking purposes. Mainly to give all the tools that a parent needs to them before closing the app.

How we built it

We worked within a Flask-Python, React-Native framework and utilized the APIs that were available in the competition. For the standard, information and task related things within the app, we just stayed within the boundaries of React-Native. For the personalized assistant, we utilized OpenAI's API to make requests and create a special assistant. For the voice cloning and reading stories feature, we used LMNT's API to create voice clones of our team members (The API only allowed one of us to clone our voices, so I volunteered haha -_- ) and make it read stories generated by OpenAI. For the article bringing, we used You.com since they had a better feature for web-scraping with queries and filters. For the emotional journaling, we used Hume.ai's API to evaluate the recorded voices in both language and prosody, and tracked them within the app. We wrapped everything in different app routes within Flask, and did the front-end with React-Native.

Challenges we ran into

For Hume.ai, we actually had an idea of training our own AI with baby-cries or baby-expressions to allow parents to communicate with their babies as well, however, we realized that there are not very useful dataset that involve infants or babies. This was a very anticipated feature for our team, and we have been talking about it since the beginning of the Hackathon. Losing this feature was quite a strike to us. Another challenge was juggling too many APIs at the same time and doing it within a Flask App on the backend. However, we overcame this by not sleeping through the Hackathon and testing it millions of times on multiple devices hahaha!

Accomplishments that we're proud of

At many points of development within these 24 hours, we thought that the project was too ambitious with too many features. However, with a few features that we needed to abandon such as cry-translator with Hume, it was overall a big success in implementing many features within a small Hackathon Window. We are proud that all the main features are working, and we may have built something that could be the start of a big thing. We went through learning 3 new APIs for us in one day, combining them in a project, building a wrapper around them, and coding a nice-enough-looking frontend that is functional with all its main features listed at the top.

What we learned

No project is too ambitious when you focus enough. And sometimes, taking chances with risky decisions to develop something that actually means something to the world might be the correct choice. Technically, we learned new APIs, gained a new App under our name, and learned that we work well together since it was our first time with this team. Generally, we had fun, and learned to have fun while coding and creating an actual app in one day. Never thought it would be possible to create this in 24 hours with immense teamwork between backend person, frontend person, and a designer. We could do it!

What's next for BabyLlama

We are pleased with the way how it shaped up in 24 hours, however, we would like to continue developing the app and make it ready for deployment. Of course, on the way, we would like add more features that we could not add due to the time limitation, such as multiple voice profiles, better UI components, an onboarding survey, maybe the features we had in mind early such as the cry-translator, and many more. We are confident of the potential of this application, tapping into this niche market with huge volume that had not been on the frontlines of AI development. If you would like to support us for future project, please contact us! It was fun developing this in 24 hours!