Swift Strategy: Accelerating Acquisitions Through AI-Assisted Strategy Development
Name
Swift Strategy IEEE Paper.pdf
Size
2.47 MB
Format
Adobe PDF
Checksum (MD5)
0b6b687a2d86c4c2212e65d5e6957187
Author(s)
Johnson, James
Burrill, Daniel
Date Issued
October 26, 2026
Abstract
Acquisition strategy development is a timeconsuming, resource-intensive process requiring the synthesis of complex guidance, technical requirements, and multidisciplinary expertise. Current manual approaches of developing acquisition strategies often span weeks or months, delaying program execution and capability delivery to warfighters. This paper introduces Swift Strategy, an artificial intelligence-driven framework that leverages multi-agent systems, retrieval-augmented generation (RAG), and human-in-the-loop workflows to accelerate acquisition strategy development while maintaining compliance and human decision authority. Swift Strategy V1 was developed and successfully deployed on the MIT TX-D2D system, demonstrating operational feasibility of AI-assisted acquisition planning tools. Building on lessons learned, Swift Strategy V2 incorporated specialized AI agents, each responsible for distinct acquisition domains including requirements analysis, acquisition pathway recommendation, financial planning, contracting strategy, risk assessment, schedule development, and more that coordinate to generate comprehensive acquisition strategies. The RAG implementation grounds all outputs in shared documents, ensuring accuracy and traceability while working to reduce AI hallucinations. Comparative analysis of multiple large language models reveals critical performance differences, with smaller models exhibiting significant hallucination rates that necessitate RAG integration. The human-AI collaboration framework enables iterative refinement where program managers provide feedback to the agents, preserving human expertise and decision authority. Swift Strategy directly addresses the Department of War (DoW) 2025 Acquisition Transformation Strategy directive to leverage AI tools for greater speed and quality in acquisition planning, offering a replicable framework for DoW-wide adoption.
MIT Department
Lincoln Laboratory
Persistent DSpace Link
