FieldCopilot: Crop lifecycle decision support for maize and rice

FieldCopilot is a crop lifecycle assistant designed to help users manage maize and rice from planning to post-harvest. It combines structured agricultural knowledge, environmental context, image-based diagnosis, and controlled information retrieval to support clearer agricultural decisions.

What is FieldCopilot?

FieldCopilot is a crop lifecycle assistant built to support maize and rice farming from planning through harvest and post-harvest. Rather than focusing only on disease detection, it is designed to help users work through multiple stages of crop production using one connected system.

It brings together crop knowledge, environmental context, image-based diagnosis, and practical follow-up guidance so users can better understand what is happening in the field and decide what to do next.

Key principle

FieldCopilot is a decision-support assistant. It is designed to support judgment, not replace agronomists, extension workers, or local expertise.

Phase 1 scope

In Phase 1, FieldCopilot is intentionally focused. It is designed around two core crops:

  • Maize
  • Rice

This narrow scope allows us to build stronger knowledge coverage, clearer limitations, and more trustworthy support. Outside these supported crops, FieldCopilot may still provide general guidance, but it should not be treated as crop-specific expertise.

How FieldCopilot works

FieldCopilot combines several layers of capability into a single agricultural workflow:

Crop lifecycle knowledge

Structured guidance across planning, planting, growth stages, nutrition, irrigation, pests, diseases, recovery, and post-harvest handling.

Context-aware reasoning

Recommendations are shaped by place, season, crop stage, and environmental conditions rather than generic text generation alone.

Image-based diagnosis

For supported crops, FieldCopilot can work alongside FieldVision outputs such as ranked predictions, confidence, and AI focus area interpretation.

Controlled information retrieval

When needed, the system can retrieve external agricultural information such as supplier or product discovery, without pretending to maintain a full commercial database internally.

What it helps with

FieldCopilot is designed to support the kinds of questions that appear throughout the crop cycle, not just after disease appears.

Planning and planting

  • Crop calendars and planting windows
  • Seed and early establishment considerations
  • Location-aware starting guidance

Growth and management

  • Soil and nutrition questions
  • Water and irrigation guidance
  • Deficiency and stress interpretation

Pests and diseases

  • Image-linked diagnosis support
  • Symptom interpretation and look-alikes
  • Follow-up management discussion

Harvest and post-harvest

  • Harvest timing guidance
  • Storage and post-harvest handling
  • Risk reduction after harvest

Environmental context

One of the most important ideas behind FieldCopilot is that agricultural intelligence should not begin with generic answers. It should begin with context.

FieldCopilot uses environmental context to determine where farming is happening, what season and conditions apply, what environmental signals matter, and which agricultural realities shape the current decision.

  • Location and agroecological setting
  • Season and crop timing
  • Weather and environmental conditions
  • Crop stage and likely risks

This helps FieldCopilot become more situationally aware and more trustworthy than a generic agricultural chatbot.

Types of questions you can ask

About the field situation

  • "What should I pay attention to at this stage?"
  • "Is this a risky time for pests or disease?"
  • "What could explain these symptoms?"

About diagnosis and symptoms

  • "Why was this disease suggested?"
  • "What look-alikes should I compare against?"
  • "What symptoms should I confirm in the field?"

About management options

  • "What are my next practical options?"
  • "What can I do first before stronger intervention?"
  • "How do I reduce this risk next season?"

About inputs and discovery

  • "Where can I look for this input?"
  • "Are there local suppliers or services?"
  • "What should I compare before buying?"

Where answers come from

FieldCopilot responses are grounded in structured agricultural knowledge and controlled system components rather than unconstrained language generation alone.

  • Structured crop lifecycle knowledge in FieldKB
  • Environmental and location context from the platform
  • Image-linked diagnostic outputs from FieldVision where applicable
  • Environmental information such as weather or climate-linked context
  • Controlled external information retrieval when needed

Transparency principle

If the system lacks sufficient basis for a precise answer, it should narrow scope, state uncertainty, or provide only general guidance rather than overclaiming certainty.

Limitations

FieldCopilot is designed to be useful, but its authority is intentionally bounded.

FieldCopilot can

  • Support maize and rice lifecycle guidance in Phase 1
  • Help interpret crop conditions and likely options
  • Combine diagnosis, knowledge, and context into one workflow
  • Support structured follow-up decision discussion

FieldCopilot cannot

  • Replace agronomists or extension officers
  • Guarantee diagnosis or outcome certainty
  • Serve as final regulatory or legal authority
  • Act as a fully autonomous farm management system
  • Provide crop-specific expertise outside supported scope with the same authority

Getting started

FieldCopilot can be entered from more than one direction depending on what the user is trying to do.

After diagnosis

Use FieldCopilot after FieldVision results to ask follow-up questions, compare options, and discuss what to do next.

As a standalone crop assistant

Start with general crop questions about maize or rice planning, management, symptoms, recovery, or post-harvest topics.