Inspiration
Non-profits constantly face challenges getting their resources and volunteers to the people that need them the most. When the physically vulnerable seek out food and shelter, sometimes they don't realize how close they are to help and housing. When the emotionally vulnerable could most benefit from a support line, they are often not in the right state of mind to make the phone call and ask for help.
What it does
Wellbot AI is a project that dynamically bridges these gaps between the needy and their needs. It does this by using an LLM fine-tuned know every non-profit program and facility in the country. Using this information, it can direct users to the closest and most appropriate source of assistance for their particular need. Additionally, user interactions will be analyzed by Hume.ai's cutting edge emotional analysis API for signs of extreme emotional distress. With this it can flag users and direct them to emotional support resources that the users' needs at that moment.
How we built it
WellBot AI was built with Hume's Empathic Voice Interface(EVI). This Interface immediately gave our project the ability to take voice communications and give us emotional metrics which we used to guide model decision making.
We expanded Hume's chatbot capabilities by adding our own tools which would help the model search the internet when a user requested a specific sort of resources.
Our front end was built using the Pygame library. We wanted to display an inviting, animated interface that users are willing to talk to.
Challenges we ran into
Learning to use Hume's API and integrating it with asynchronous tools such as Pygame. At the start, we had trouble translating the AI's words onto a fun, presentable screen seamlessly. Since Hume is on the cutting-edge, we had to manage the correct Python versions and Hume versions to make sure that it would work without error.
Accomplishments that we're proud of
We are proud of implementing our own tools with Hume to search specific databases and use specific queries. We were also able to use OpenAI's tools in Hume to enhance the performance and capabilities of our AI chatbot.
What we learned
We became familiar with Hume's API and other popular APIs that will be used to strengthen the foundation of our AI knowledge. We learned the importance of tools in AI and deeply understanding them to ensure maximum efficiency.
What's next for WellBot AI
The next step for WellBot AI is to fine-tune our chat bot and expand the range of services it knows and regions it can be active in. Many food banks, shelters, and other support services are local initiatives that are tailored for regional needs. These resources must be known to our AI so when users in any region ask for help, it will be ready to provide it. Large disasters that strike our communities can utilize this product to get help where it's needed quickly.
Log in or sign up for Devpost to join the conversation.