HarvestIQ: AI-Powered Post-Harvest Intelligence

Helping farmers reduce post-harvest food waste by connecting produce to timely market opportunities with AI-powered intelligence.

Comment

About the project

The problem

A farmer can successfully grow a crop and still lose money because the crop cannot reach a buyer in time.

In many agricultural communities, farmers have limited access to timely market information. They may not know the current selling price, where demand exists, or which buyers are actively looking for their produce. As a result, harvested crops can remain unsold until they deteriorate.

This is not only an economic problem. Food that is produced but never consumed represents wasted land, water, fertilizer, energy, labor, and transportation resources. Reducing post-harvest loss is therefore both an agricultural and environmental challenge.

What inspired HarvestIQ

HarvestIQ was created around a simple question:

What if a farmer could know what to do with their harvested produce before it becomes waste?

We built HarvestIQ as an AI-powered post-harvest intelligence platform designed around the realities of smallholder farmers, including farmers who may have basic mobile phones rather than smartphones or reliable internet access.

Instead of requiring a farmer to install an application, HarvestIQ provides access through USSD and SMS while also providing a web interface for buyers and other users.

What we built

HarvestIQ connects farmers, market information, buyers, and AI-assisted decision-making into one system.

A farmer can use USSD to:

  • Check current produce prices.
  • List available produce for sale.
  • Provide information about their harvested produce.
  • Access AI-powered agricultural guidance.

Buyers can use the platform to discover available produce and post demand.

Behind these interactions, HarvestIQ maintains market and farmer data and uses automated processes to keep information updated and generate relevant alerts.

The goal is not simply to create another agricultural marketplace. The system is designed to help farmers make time-sensitive post-harvest decisions that can prevent produce from becoming unsellable.

How the system works

Farmer
   │
   │ USSD / SMS
   ▼
HarvestIQ Platform
   │
   ├── Market Prices
   ├── Produce Listings
   ├── Buyer Demand
   └── Farmer Information
           │
           ▼
      AI Intelligence
           │
           ▼
   Actionable Recommendation
           │
      ┌────┴────┐
      ▼         ▼
   Farmer      Buyer
    SMS        Platform

For example, instead of a farmer discovering several days later that their produce has lost market value, HarvestIQ can help surface relevant market information and buyer demand earlier, giving the farmer an opportunity to sell or redirect the produce before spoilage occurs.

Technology

HarvestIQ is built as a cloud-based serverless platform using:

  • Node.js for backend services.
  • AWS Lambda for serverless computation.
  • Amazon DynamoDB for farmer, market, listing, demand, and alert data.
  • Amazon EventBridge for scheduled market-data and alert-processing workflows.
  • AWS Bedrock for AI-powered agricultural intelligence.
  • Africa's Talking for USSD and SMS communication.
  • A web dashboard for buyers and platform interaction.

The architecture allows the platform to remain lightweight while supporting users who may have limited connectivity.

What we learned

Building HarvestIQ taught us that solving a real-world problem is not simply about adding more technology.

One of our biggest lessons was the importance of designing around the user's actual constraints. A smartphone application may be technically impressive, but it does not solve the accessibility problem for a farmer who owns a basic phone or has unreliable mobile internet.

This led us to make USSD a central part of the platform rather than treating it as an optional feature.

We also learned how different technologies have to work together to solve a practical problem: communication infrastructure, cloud computing, databases, automation, AI, and user-facing interfaces all have to form one reliable workflow.

Challenges

One of the biggest challenges was designing an AI-enabled system that remains useful without requiring constant internet access from the farmer.

Another challenge was connecting asynchronous cloud services, market information, USSD interactions, alerts, and AI processing into a coherent workflow.

We also had to think carefully about how to present AI recommendations so that they support the farmer's decision rather than simply producing generic AI-generated text.

Why it matters

HarvestIQ approaches food waste from the point where an intervention can still make a difference: after harvest but before spoilage.

Every kilogram of produce successfully sold or redirected before deterioration represents not only value recovered for a farmer, but potentially food that does not become waste after the resources required to produce it have already been spent.

Our long-term vision is to develop HarvestIQ into a broader post-harvest intelligence infrastructure that combines market intelligence, AI, connectivity-independent access, and eventually physical sensing technologies to help farmers reduce losses and build more resilient agricultural systems.

HarvestIQ's mission is simple: help more of what farmers grow reach the people who need it.

Built With

Share this project:

Updates