Everside Companion

Everside Companion app with User Language English
Everside Companion app with User Language English
Everside Companion app with User Language English

Inspiration The main objective of an AI-enabled senior companion app is to provide support, companionship, and improved well-being for older adults. The app can assist them with daily tasks, medication reminders and information access, promoting greater independence for seniors.

What it does

The app can provide conversation, entertainment, and a sense of connection and combating loneliness . By taking care of the daily needs and providing social connection, the app can contribute to a better quality of life for seniors.

How we built it

The application was build on Amazon PartyRock which is a free hands-on environment for experimenting with generative AI. PartyRock allows for text generation, image generation, and ChatBot capabilities.
We described the features of the application through the prompt provided and the platform created the respective widgets.
After this we created multiple Users widgets to collect the user details to make it more personalised and provide recommendation based on the user preference. For eg. in this case we provided, their interest, location, languages, food preferences etc.
The generated app gave custom generative widgets with LLM by using various attributes provided in we can customise these widgets and also mapped to the user inputs. For eg. the user preference food was mapped to grocery, user interest mapped to interest and recommendation etc.,
Customizing prompts, adjusting widgets and exploring ChatBot option we could come up with an interactive contextual assistance to the Elderly person.

Challenges we ran into

While rendering more AI Powered widgets using a single dependent user input widget, there were AI response failures Defining relationship between widgets and user input fields. Enable to upload the image, doc or excel. Fine tuning system prompts to generate the AI response uniquely and consistently. Image generations are not working as easy as we do integration.

Accomplishments that we're proud of

Finally, we are very happy, because we could do the testing with multiple user preferences and get customized details as output from the chat. A personalised app that is highly customisable and generates dynamic data to represent accurate data. One solution, that covers most of the needs for the target audience.

What we learned

How to efficiently write prompts to get the maximum details from Partyrock AI. How to provide Prompt chaining. Retrieval Augmented Generation based on user inputs.