Curable

• 

  App UI

Inspiration

As an international student juggling numerous responsibilities, I often find it challenging to focus on my health and well-being. Monitoring and managing my health effectively has been difficult. A recent medical experience underscored the importance of robust health management and highlighted significant gaps in my approach. Through this episode, I discovered that gaining a deeper understanding of health challenges not only increased my confidence but also showed me how an informed approach could empower me to manage them more effectively. This realization highlighted the need for a centralized platform to streamline health management tasks, providing personalized insights and resources to help individuals maintain optimal well-being.

What it does

This agent targets individuals with chronic health conditions. It offers functionalities such as analyzing medical reports, logging symptoms, providing information about conditions, managing nutrition, and general health education. It logs meals with nutritional data and retrieves daily caloric intake. Additionally, it analyzes health metrics like heart rate and steps and provides suggestions for improvement. The main goal is to simplify and enrich the user's healthcare experience.

How it is built

The application is built around specialized sub-agents, each tailored to manage specific tasks efficiently. These sub-agents are seamlessly integrated with a default main agent that orchestrates overall functionality and invokes the appropriate sub-agents. Some sub-agents are connected to data stores, enabling interaction with unstructured data stored in cloud buckets, while others interface with cloud functions that link to Firestore Databases. This architecture ensures that user interactions are not only well grounded but also specific to user's unique circumstances.

Agents Design

Challenges

Challenges included refining agent invocation and testing the agents with and without integrating few-shot examples to enhance the agent's accuracy and responsiveness.

Accomplishments

The integration of multiple agents to work seamlessly together, providing accurate and useful information to users. Another significant accomplishment is demonstrating the ease of building such a system with low-code solutions provided by GCP as seen below.

App Design

Learnings

This project enhanced my understanding of AI and conversational interfaces, demonstrating how important is to effectively train agents/models. This experience has sparked a deeper interest in exploring more features within Vertex AI.

What's next for Curable

Future development will focus on expanding backend functionalities to handle more users. Plans include implementing voice integration to enhance accessibility and user engagement.

Built With

• cloudfunctions
• firebaseauth
• firestore
• python
• react