Inspiration
The inspiration behind the Agentic App for Vegetable Harvesting stems from the desire to simplify gardening and farming decisions using advanced technology. As a software developer and an avid backyard farmer, I recognized the need for a smart solution that could provide timely and accurate recommendations for sowing and harvesting vegetables. This project not only enhances productivity in gardening but also serves as a learning platform for leveraging advanced AI and graph databases.
What it does
The Agentic App acts as a Smart Gardening Advisor, guiding users on when to sow and harvest a wide variety of vegetables. By integrating Large Language Models (LLMs) through LangChain and combining it with a graph database (ArangoDB) and NetworkX analytics, the app can: Provide sowing and harvesting schedules based on user queries. Suggest the best vegetables to plant in a particular month. Identify the month with the largest variety of vegetables to sow or harvest. Answer advanced agricultural questions using dynamic query routing.
How we built it
Data Conversion to Graph: The dataset was converted into a graph where nodes represent months and edges indicate which vegetables can be sown or harvested between months. This graph was initially created using NetworkX and then persisted into ArangoDB.
Hybrid Query System: Implemented a Hybrid Query Router to dynamically select between AQL (ArangoDB Query Language) and NetworkX based on the complexity and type of user queries Tool Integration:
The project utilizes: ArangoDB AQL Tool: Executes AQL queries on the stored graph for precise data retrieval. ** NetworkX Tool: ** Runs Python-based graph algorithms to handle advanced analytical queries. *LLM Integration: * Uses LangChain to generate, interpret, and execute queries using GPT-4, ensuring natural language understanding and responses.
Challenges we ran into
Data Consistency: Ensuring the graph data remains consistent when switching between NetworkX and ArangoDB tools. Dynamic Query Routing: Building a reliable system to automatically select the most appropriate tool for each query. LLM Integration: Fine-tuning the model to generate accurate and contextually relevant AQL and Python code.
Accomplishments that we're proud of
Successfully implemented a smart query router that seamlessly switches between AQL and NetworkX based on the query. Achieved robust integration of LangChain with a hybrid graph analytics approach. Created a practical and user-friendly app that provides valuable agricultural insights using state-of-the-art AI tools with precision.
What we learned
How to effectively use LangChain with LLMs to process natural language queries. The power of ArangoDB for graph storage and advanced AQL queries. Leveraging NetworkX for performing complex graph analytics directly on Python-based graphs. Building a dynamic, hybrid query system that can automatically adapt to different query types and complexities.
What's next for Agentic App to harvest vegetables
Expand Dataset: Integrate seasonal and regional data to provide hyper-localized gardening advice. User Interface: Develop a mobile and desktop-friendly interface to make the app more accessible to everyday gardeners. Advanced Analytics: Introduce predictive analytics for future harvest planning. Integrate IoT: Connect with smart garden sensors to provide real-time advice based on environmental data. Enhanced AI Models: Fine-tune the AI model further for even more accurate and insightful gardening guidance.
Log in or sign up for Devpost to join the conversation.