CropCompanion

• 

  farm data input

• 

  chat page

• 

  signin page

• 

  backend

Inspiration

As the world increasingly turns to data for decision-making, we realized that the backbone of the food chain—the farmers—were being left behind, lacking access to actionable data-driven insights. With the goal of helping farmers maximize their crop yields sustainably, we set out to provide them with the tools they need to make informed decisions.

What it does

CropCompanion is an innovative application designed to empower farmers and gardeners by providing them with personalized insights and recommendations to optimize crop health and yield. The app leverages data from various sources to help users make informed decisions about their crops, taking into account local conditions such as weather, soil health, and other essential factors.

How we built it

• On-Device LLM: We used Llama 3.2 for on-device language model processing, enabling farmers to interact with the app and get insights without needing to understand complex technical details.
• MongoDB: We used MongoDB to store the messages and ensure the app retains memory of the ongoing conversation, allowing the language model to reference previous exchanges and provide more personalized advice.
• Ollama for LLM Server: Ollama was used to run the LLM server, enabling seamless processing of data and responses for farmers.
• Backend with FastAPI: The backend of the application was built using FastAPI, ensuring high performance and rapid responses for users.
• Frontend with Next.js: The frontend was developed using Next.js, offering a fast, responsive, and intuitive user experience.

Challenges we ran into

• Providing Memory for the LLM: One of the primary challenges was giving the LLM memory for the current conversation, ensuring it could retain context throughout interactions with the user.
• Building the AI Model: Developing a reliable AI model to predict which crops would thrive based on parameters like weather, soil type, and other factors proved to be a complex task.
• Integrating All Modules: Coordinating the integration of different modules across the team was challenging, but ultimately we managed to connect the backend, LLM, and front-end seamlessly.
• Enabling Authentication: Implementing a secure authentication system to protect user data and provide personalized experiences required careful planning and execution.

Accomplishments that we're proud of

• Successful MVP: We are proud to have successfully created a minimum viable product (MVP) that we can pilot with real farmers, providing valuable insights into crop management. Solving Memory and Context Issues: Overcoming the challenges of memory and context for the LLM was a major accomplishment, enabling the app to provide more coherent and personalized advice.
• Authentication Setup: Setting up robust authentication was another milestone, ensuring the security and privacy of our users.
• Team Coordination: The collaboration between our team members was key to pulling everything together and making the app a reality.

What we learned

• Data Management and Memory: We learned the importance of managing and retaining data to improve the quality of AI responses. Contextual memory was essential for creating a natural, coherent conversation flow.
• Collaborative Problem-Solving: Building a project with multiple technical components taught us how to work efficiently across disciplines and solve challenges together as a team.
• AI in Agriculture: We gained a deeper understanding of how artificial intelligence can be applied to agriculture, helping farmers leverage technology to improve crop yields and sustainability.

What's next for CropCompanion

• Expand the Crop Prediction Model: We aim to refine and expand our crop prediction AI model, incorporating more data sources and machine learning techniques to increase the accuracy of our recommendations.
• User Feedback Integration: As we pilot the app with real farmers, we plan to gather user feedback and continuously improve the app based on their experiences.
• Mobile App Development: We are planning to develop a mobile version of CropCompanion to make the app more accessible to farmers on the go.
• Weather Data Integration: We aim to enhance the app by integrating real-time weather data, helping farmers make immediate decisions based on current conditions.
• Sustainability Focus: Moving forward, we plan to deepen our focus on sustainable farming practices by providing more insights into eco-friendly farming techniques and resource management.

Built With

• fastapi
• llama3.2
• llm
• mongodb
• next.js
• oauth2
• ollama