Skip to main content

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Share your experience with the FAS IT-Playbook by taking this brief survey

AI Solutions

The PPMS system is embarked the journey of implementing AI-driven solutions, streamlining personal property utilization & donation and optimizing sales strategies. These AI implementations enhance efficiency, reduce costs, and improve decision-making, leading to better outcomes for taxpayers and stakeholders.

Introduction

AI is being incrementally adopted within GSA PPMS through targeted, scalable use cases. This outlines the architecture and enabling technologies supporting AI across PPMS and GSA Auctions, including production and proof-of-concept implementations using platforms such as Spring AI and Amazon Bedrock, with controlled, low-risk implementation and appropriate human oversight.

Use Case 1: Automated Lot Description Generation for Auction Items – In Production

The GSA Auctions platform requires high-quality lot descriptions to streamline the auction listing process and improve buyer engagement. In GSA Auctions, a lot represents an auction event that contains up to 999 items. Manually creating comprehensive, consistent, and compelling descriptions for lots is time-consuming and prone to inconsistency. This use case leverages Spring AI integrated with Amazon Bedrock's Claude 4 to automatically generate professional, detailed lot descriptions based on structured item data and supplementary visual information.

Technical Architecture

Architecture diagram titled 'PPMS VPCaaS Account' showing a Spring AI integration flow within an AWS environment. Inside the PPMS VPC, a PPMS UI connects to a PPMS API, which in turn reads from and writes to a MySQL PPMS RDS database. The numbered data flow proceeds as follows: (1) the PPMS API sends Raw Input into the Spring AI processing layer; (2) a combiner merges the Raw Input with a Prompt to produce Input + Format Prompt Instructions; (3) those instructions are sent to Amazon Bedrock in a separate FCS AI Account; (4) Amazon Bedrock returns Raw Output; (5) the Raw Output is passed to a Structured Output Converter; and (6) the Structured Output is returned to the PPMS API.