Don Patio: Post-Harvest Decision Agent for Fruit Producers

Inspiration

Don Patio was inspired by the reality that many fruit producers lose a significant portion of their harvest after it has already been collected. These losses are rarely caused by poor farming practices, but by uncertainty and delayed decisions during the post-harvest stage. Producers often need to act quickly with limited information, making decisions based on intuition rather than structured analysis. The project emerged from the idea that AI could serve as a practical decision partner during these critical moments.

What it does

Don Patio is an AI-powered post-harvest decision agent for fruit producers. It evaluates spoilage risk, urgency, and recommends concrete actions based on real-world conditions such as fruit type, ripeness, volume, storage conditions, time since harvest, and perceived climate. Instead of offering generic advice, Don Patio delivers clear, actionable recommendations and explains the reasoning behind each decision.

How we built it

The project was built using Google AI Studio and the Gemini 3 API. Don Patio was designed as a decision-oriented agent that reasons over multiple variables provided by the producer. Gemini 3’s enhanced reasoning capabilities enable the agent to assess trade-offs between time, storage, and product condition, generating low-latency responses suitable for real-world agricultural use.

Challenges we ran into

One of the main challenges was preventing the agent from producing overly generic responses. Ensuring that each recommendation reflected true reasoning rather than predefined rules required careful prompt design and iteration. Another challenge was building a meaningful solution without relying on external APIs or real-time data, which pushed the project to focus on reasoning quality and explainability.

Accomplishments that we're proud of

We successfully built a functional AI agent that goes beyond conversational assistance and supports real-world decision-making. Don Patio demonstrates how Gemini 3 can be used to create explainable, decision-oriented agents with tangible impact. The project shows that meaningful agricultural tools can be built with minimal infrastructure when reasoning is placed at the center of the design.

What we learned

We learned that large language models are most powerful when applied to decision-making rather than information retrieval. Gemini 3’s ability to reason across multiple variables and generate transparent explanations is critical for building trust with users in non-technical environments. We also learned the value of simplicity when designing tools for time-sensitive contexts.

What's next for Don Patio: Post-Harvest Decision Agent for Fruit Producers

Future iterations of Don Patio will integrate external signals such as market trends, weather data, and storage optimization models. The long-term vision is to evolve Don Patio into a comprehensive post-harvest intelligence layer that helps producers reduce waste, improve planning, and strengthen their connection to markets.

Built With

• ai
• gemini
• google-directions