The Center for Data Innovation recently spoke with Mario Bustamante, CEO of Instacrops, a Chile-based agricultural-technology company developing AI agents that help farmers use real-time field data to make growing decisions. Bustamante explained how Instacrops combines data from sensors, satellites, and weather systems to provide recommendations on irrigation, crop management, and other agricultural decisions.
David Kertai: What does Instacrops offer?
Mario Bustamante: Agricultural technology has become increasingly sophisticated, from soil-moisture sensors that track conditions hour by hour to satellite imagery that maps crop stress across entire fields, but farmers still struggle to turn that growing stream of data into decisions they can act on. A farm may generate thousands of data points a day, but those numbers do not answer the questions farmers actually face, such as whether a field needs water, whether incoming weather could damage a crop, or whether changing soil conditions could affect yields.
These decisions can shape an entire season, and farmers make them in rural environments where connectivity can drop, hardware must operate in harsh conditions, and crop and soil conditions can change quickly. Under these conditions, a month-long or year-long plan can break down when the field behaves differently than expected.
Instacrops tackles this problem by using AI agents to turn agricultural data into practical recommendations. We combine regional and on-site information from field sensors, satellite imagery, and weather systems to help farmers make decisions about irrigation, weather risks, crop management, and yields. One of our most widely used agents, InstaDrop, evaluates soil moisture, weather forecasts, and crop requirements to recommend when farmers should irrigate. We deliver these recommendations through WhatsApp, allowing farmers to receive guidance, ask questions, and share what they see in the field in real time.
Kertai: How does Instacrops collect data from fields?
Bustamante: We deploy thousands of sensors across farms in Latin America that measure soil moisture, temperature, humidity, rainfall, solar radiation, and wind. We combine these measurements with satellite imagery, weather forecasts, historical farm data, crop information, and other external sources. Farmers can also send our AI agents text messages, voice notes, photos, and observations through WhatsApp.
Together, these sources give our AI agents a more complete picture of each field. The system interprets these signals in context. For example, a soil sensor may show moisture dropping quickly while a weather forecast predicts rain later in the day. Combining those signals helps the system account for changing conditions when generating recommendations.
Kertai: How do your AI agents turn collected data into recommendations?
Bustamante: Our system combines farm-specific information with an AI system and agronomic knowledge to analyze what is happening in a field. Specialized AI agents apply that context to specific tasks. For irrigation, the agent considers soil moisture at multiple depths, recent irrigation events, weather conditions, forecasts, crop stage, and water requirements before determining whether a field needs water.
We deliver recommendations through WhatsApp, where farmers can ask follow-up questions or describe what they are seeing. If a farmer sends a photo of standing water or reports that a pump malfunctioned, the system can incorporate that information immediately. Instead of simply displaying data, the technology uses it to help farmers decide what to do next.
Kertai: How do you ensure the accuracy of your AI agents’ recommendations?
Bustamante: We ground recommendations in farm-specific data, agronomic rules, historical information, and real-time measurements. We validate our models with agronomists and compare recommendations with actual field outcomes.
The AI agents also account for uncertainty. If key information is missing, such as incomplete soil-moisture data or a sudden change in weather, the system identifies the gap and requests additional context instead of making a guess. We continuously compare what the system recommends with what happens in the field and use those results to improve its performance.
Kertai: Could you share any use cases of Instacrops’ technology?
Bustamante: Irrigation is one of our strongest use cases. Across deployments, our AI agent InstaDrop has helped farmers reduce water use by around 30 percent on average while maintaining or improving crop performance. In regions where water availability is becoming a critical constraint, that impact is significant. We are also applying the same approach to frost and weather risk, yield forecasting, crop monitoring, and nutrition. The goal is to help farmers use the data already available to them without requiring them to become data scientists.
