FurWell: Your 24/7 AI Pet Health Assistant

Prototype1: app's vision
Prototype2: app's vision

Inspiration

As pet owners, the three of us have all faced heart-wrenching moments: your pet starts acting unusual, but the vet is far away, closed, or simply unavailable. Angela remembers a sleepless night worrying over Jessica, her British Shorthair, who was limping mysteriously, and every Google search only added to her confusion. Diane once rushed her cockatiel, Lio, to an avian specialist hours away, terrified by sudden coughing sounds, only to learn it was a minor issue. Xinwei recalls a similar moment of panic when his Samoyed, Mango, couldn’t stop pacing and whining late at night, with no vet open nearby. These moments of uncertainty stayed with us. They showed us a clear gap between the knowledge we need as pet owners and the access we have to immediate veterinary advice. We asked ourselves: what if there was a way to provide instant, reliable guidance that could bridge the gap, offering clarity in those critical moments? That’s how FurWell was born. Inspired by the challenges we’ve all faced, we set out to create an AI-powered assistant that delivers instant, actionable insights for pet health. More than just identifying what might be wrong, FurWell is designed to guide you on the next steps with confidence, using a trusted veterinary knowledge base as its foundation. Through FurWell, we hope to turn moments of panic into moments of clarity for pet owners everywhere - whether you’re caring for a curious cockatiel, a gentle British Shorthair, or an energetic Samoyed. It’s not just an app - it’s the tool we wish we had in those sleepless, anxious hours.

What it does

Our app is an AI-powered chatbot designed for pet owners, especially cat and dog parents, to provide pre-diagnostic advice and actionable suggestions based on symptoms or conditions. Backed by authoritative veterinary data, the app empowers pet parents to make informed decisions about their pets’ health, whether to manage symptoms at home or seek professional veterinary care.

Key Features

AI-Powered Chatbot

Users can ask questions about their pets’ health or behavior, and the chatbot provides responses based on veterinary-grade data.
It helps users understand potential issues and offers guidance on next steps.

Multi-Pet Management

Users can add multiple pets to the app and manage their profiles individually.
Each pet has its own record, enabling users to keep interactions and health data organized.

Clinical History Tracking

The app allows users to log and save their pets’ medical history, including vaccinations, past illnesses, and treatments.
This feature aids in better tracking of long-term health trends and provides context for addressing current concerns.

Daily Activity Logs

Users can record daily updates about their pets, including activities, diet, mood, and health conditions.
This feature helps users monitor patterns and detect potential health issues early.

Future Vision: Community Building

While the current version doesn’t include a community feature, the app is designed with plans to foster a community for pet owners.
This future functionality will allow users to share experiences, tips, photos, and videos, creating a supportive and informative space for pet parents.

How we built it

Data Acquisition：Sourcing comprehensive veterinary data to inform and empower our AI-driven system.

Since free veterinary datasets were unavailable, we turned to trusted animal health textbooks as our main source of information. These textbooks provided a reliable foundation, covering essential topics like medical conditions, treatment approaches, and species-specific care. By focusing on high-quality and well-regarded resources, we ensured our AI system has access to accurate and practical veterinary knowledge, setting the stage for trustworthy recommendations for pet owners.

Implementing Retrieval-Augmented Generation (RAG)： Enhancing AI responses with precision-driven retrieval mechanisms.

We developed a strategy to improve the relevance and quality of AI-generated responses by incorporating Retrieval-Augmented Generation (RAG). This involved combining data augmentation, query optimization, and Cortex filter search with smart metadata matching. Key improvements included:

Making Inputs Clearer: Refining user queries to align better with our database and improve specificity.
Adding Context: Using information from the user’s input, past interactions, and metadata to enrich the AI’s understanding.
Targeted Filtering: Delivering more accurate results by considering factors like species type, symptom severity, and urgency. This process created a robust system that consistently delivers actionable insights to pet owners. It forms the backbone of FurWell’s AI recommendations, helping users make informed decisions about their pets’ health.

Building and Testing the RAG with Large Language Models (LLMs)：Ensuring accurate, contextually relevant, and user-centric AI responses.

To build a reliable RAG pipeline, we used advanced Large Language Models (LLMs) and ran extensive tests to ensure the system met real-world needs. This included:

Cortex Search and Query Refinement: Aligning Cortex Search with clearer, more structured queries for better results.
Metadata-Based Filtering: Tailoring responses by incorporating information like pet type, condition severity, and urgency.
Scenario Testing: Simulating a wide range of pet health issues to validate the accuracy and usefulness of the AI’s recommendations. Through these efforts, we created an AI system that pet owners can trust. It provides clear, empathetic, and actionable advice, filling a crucial gap for those seeking immediate guidance for their pets.

Developing the Frontend：Creating a seamless and user-friendly interface using Streamlit.

We designed a frontend using Streamlit to make the AI system easy to use and accessible for everyone. Key features include:

Easy Navigation: A simple layout that lets users input queries without hassle.
Interactive Tools: Real-time responses and dynamic forms to keep users engaged.
Personalized Results: Options to customize advice based on a pet’s unique needs. Our user-first design approach ensures that pet owners can quickly get the information they need in a way that feels straightforward and intuitive. This interface bridges the gap between advanced AI and everyday users.

Challenges we ran into

Limited Access to Veterinary Data
- Finding high-quality, authoritative veterinary data was challenging, as much of it is privately owned or inaccessible to the public. This made sourcing reliable information time-intensive.
Complexity of Snowflake Documentation
- Snowflake’s documentation was intricate and difficult to navigate, especially for first-time users, slowing down our data integration process.

Steep Learning Curve with New Technologies

Our team had to familiarize ourselves with several new technologies, from backend frameworks to AI deployment tools, which required extensive learning during the hackathon.

RAG Strategy Iterations

Implementing Retrieval-Augmented Generation (RAG) required us to experiment with multiple strategies, fine-tuning approaches to identify the most efficient and accurate method for integrating authoritative data.

Accomplishments that we're proud of

Successfully Built the App
- Despite the time constraints and challenges, we successfully developed a functional app that meets the primary goals of the project.
Leveraged Real User Data
- We incorporated real user data to craft context-aware prompts for the AI model, significantly improving the chatbot’s accuracy and relevance.
Implemented Multi-Pet Management
- Adding functionality for users to manage multiple pets and their unique profiles was a significant technical achievement, enhancing usability.
Enabled AI-Powered Pre-Diagnostics
- By integrating RAG and fine-tuning the model, we delivered pre-diagnostic suggestions that are both accurate and user-friendly.

What we learned

Data Accessibility is Key
- The importance of securing access to reliable, open-source data cannot be overstated, especially when building applications that rely on authoritative knowledge bases.
Iterative Problem-Solving for RAG
- Experimenting with various RAG strategies taught us how to balance retrieval accuracy with model performance, a skill we can apply to other AI-powered applications.
Importance of Team Learning
- Tackling unfamiliar technologies pushed us to rapidly learn and adapt, fostering stronger collaboration and problem-solving skills among team members.
User-Centric Design Matters
- Through testing with real users, we learned the value of designing with empathy, ensuring the app is both functional and easy to use for pet parents.
Community Potential is Enormous
- Engaging with pet owners made us realize the demand for a platform where users can connect, share, and learn from one another, shaping the app’s long-term vision.

What's next for FurWell: Your 24/7 AI Pet Health Assistant

Our vision for FurWell extends far beyond its current features, with ambitious plans to enhance its utility and foster a vibrant community of pet owners.

Collaborating with Veterinary Experts
- We aim to partner with licensed veterinarians and veterinary organizations to gain access to first-hand clinical data. This collaboration will enable us to expand and refine our database, ensuring that the chatbot provides even more authoritative, accurate, and trustworthy advice for pet parents.
Creating a Pet-Centric Community Platform
- We envision FurWell evolving into a dedicated social space where users can share their pet stories, advice, photos, and videos. This community will provide an opportunity for pet parents to connect with one another, exchange tips, and celebrate their pets, creating a supportive and resourceful environment tailored for pet lovers.
Integrating Appointment Booking for Pet Clinics
- Future iterations of FurWell will include a streamlined appointment booking feature, allowing users to schedule vet visits directly through the app. This integration will remove the hassle of searching for and contacting clinics, making it easier for users to prioritize their pets’ health needs.
Advancing the AI-Powered Chatbot
- We plan to make the chatbot even smarter by incorporating predictive analytics, allowing it to suggest preventative care routines and identify health risks based on logged patterns and clinical data.
Enhancing User Experience through New Features
- Additional features such as health tracking dashboards, personalized calendars, and pet activity monitoring will be refined to deliver a seamless user experience. These tools will help pet owners keep track of their pets’ well-being while maintaining all essential records in one place.
Building for Global Accessibility
- We aim to expand FurWell’s reach by supporting multiple languages and offering region-specific veterinary advice to make the app accessible to pet owners worldwide. Through these future enhancements, FurWell is committed to becoming not just a health assistant but a holistic platform that supports every aspect of pet parenting.