Transforming Public Sector Procurement with Agentic AI on AWS

This article explores how agentic AI architecture on AWS modernizes public sector procurement from solicitation to proposal evaluation while maintaining compliance with FAR, DFARS, and Canadian procurement frameworks.

• Intelligent agents autonomously handle regulation-heavy tasks with human-in-the-loop oversight at every decision point
• Four-layer serverless platform uses Amazon Bedrock AgentCore, Lambda, OpenSearch, and DynamoDB for scalable multi-agent workflows
• RFx Builder workflow automates six-stage solicitation generation including requirements analysis, market research, compliance mapping, and document assembly
• Evaluation Builder provides five-stage assessment pipeline combining AI-powered scoring with human oversight and competitive intelligence
• RAG architecture with four specialized knowledge bases enforces citation requirements to eliminate hallucination in regulated environments
• Seven connector types enable data discovery across REST APIs, S3, ODBC, Databricks, Snowflake, and legacy on-premises systems
• Defense-in-depth security includes zero-trust networking, least-privilege IAM, encryption at rest, and multi-AZ high availability

The architecture demonstrates how agentic AI reduces manual effort, improves compliance accuracy, and accelerates procurement cycles while maintaining explainability and accountability through structured reasoning traces.