A12 Single Source Justification.pdf

Single Source Justification United States Department of Agriculture

FAR 13.106-l(b)(l)- Simplified Acquisitions

1. Identification of the agency and the contracting o ice.

Requesting Agency:

USDA AMS LP Livestock, Poultry, & Grain Market News 1400 Independence Ave, SW Room 2619-S, STOP 0252 Washington, DC 20250-0252

2. Description of supply or service required to meet the agency's needs.

This sole source justification recommends continuing our AI Project to grade/evaluate Feeder Cattle with Ensemble Government Services, LLC.

3. The rationale or circumstances for limiting competition.

The AI project for grading feeder cattle involves several technical stages that are time-intensive, including Data Capture, Data Annotation, Model Training, and hardware/software evaluation.

Initially, this project set out as development of an App to be used by the public to evaluate their own livestock. Two qualified vendors were part of the current Proof of Concept.

Avatar Partners Inc used iPads to capture video of feeder cattle, developing a basic app for object detection, skeletal points, and measurement, but lacked critical grading components (frame, muscle, and flesh scores). The project lasted 6 months, with a team inexperienced in agriculture.

Ensemble Government Services, LLC employed HD and LiDAR cameras, capturing more data points and focusing on essential grading components like frame and muscle scores. Their project spanned 12 months, with team members experienced in agriculture. Ensemble is also a SBA 8(a) Certified Small Disadvantaged Business.

Given the significant and rapid progress made on this project by Ensemble Government Services, LLC, and given the overall importance of this project to meet USDA’s initiative to enhance service e iciency through innovation, USDA recommends continuing its partnership with Ensemble Government Services, LLC. Transitioning to a new vendor at this critical juncture in the project would cause delays and cost overruns as a new vendor would require time to understand the project's specific requirements, capture and annotate new data consistently, and re-train the AI models from scratch. This would extend the project timeline significantly and would compromise the quality and accuracy of the AI grading system and the financial and time investments made by USDA to date.

4. A description of the market research conducted and the results or a statement of the reason market research was not conducted.

In early 2024, the USDA issued an RFI to gather information from vendors on developing an app for non-contact visual livestock measurements using technologies like LiDAR or 2D-to-3D mapping. After reviewing submissions and holding formal discussions, the USDA issued a Request for Proposal for a 1-year Proof of Concept. Multiple vendors submitted proposals on how they would proceed to show proof of concept. From those responses, two vendors were selected to provide di erent approaches and solutions.

5. Estimated price, including all options, if applicable.

$425,000

6. Recommended source(s).

Ensemble Government Services, LLC.