Produce Pics - AI Agents for Reducing Food Waste

Hyperspectral Imaging

Hyperspectral imaging will be available on smartphones within the next eighteen months.

AI Agents

Automated VLM based agents automatically download regulatory requirements from the Dept. of Agriculture site.

Shelf Life Models

The simple to use image processing platform provides features for estimating the remaining shelf-life of produce.

Global Image Database

With each captured image, the platform become more robust by capturing variations across growth regions, seasons, etc.

Farmers

Farmers will be able to leverage simple smartphones to efficiently meet market demands for higher quality produce.

Distributors

Images provide more reliable data for transparently tracking and inspecting food across the supply chain.

Consumers

Using images, consumers can order trusted and certified food with different remaining shelf-lives.

Frequently Asked Questions

Does one need new (and expensive) equipment?

The strength of the system is its simplicity - it is accessible via a smartphone and does not require any additional equipment. Pictures may be taken on the farm, uploaded later from the pack-house and still be able to trigger a PACA (and blockchain) based smartcontract.

Will the system work in different countries?

The system can comprehend regulatory text written in different languages and create inspections compliant with regulations in different countries. The app is uninversal and meets regulatory requirements in different countries.

Will the cost of Quality / USDA Inspections be reduced?

By eliminating the need for a certified inspector to travel and by including a robust image processing platform, we anticipate the cost of inspections to go down from around US $ 250 per inspections to around $ 25 per inspection.

Does one require stable WiFi connections at the Farm?

Most field areas in a farm do not receive good WiFi signals. The app allows a grower to take pictures on a smartphone first and then upload later from the office, etc.

Why use AI Agents?

Inspection guidelines for each commodity (a fruit, vegetable, beef or pork) consists of at least five large PDF documents with some reference images. LLMs (with RAG) can process them more efficiently and auto configure foundation vison models to carry out the required inspections. to

Will the wait times for Quality / USDA Inspections be reduced?

Travel time and the need to drive to multiple facilities contribute to large wait times, often in the range of a few hours to upto 25 hours. Use of the roubst image processing platform reduces wait times for inspections.

Reducing Food Loss

Roughly, one third of the food produced in the world goes to waste. The founding team has been using imaging (including hyperspectral imaging) technologies to address food loss and wastage. After extensive research with growers, distributors, and retailers, the team realized that the LLM integrated AI Agents framework combined with latest foundation vision models can help and are in the process of developing such a system.

Produce Pics was incubated at the Columbia University / IBM Launch incubator and was one of the three companies selected for non-dilutive follow on funding.

More info on Machine Learning and Sustainable Food Supply.