anywhere from 70-300 days, depending the type of contract type and methodology used. While human oversight remains irreplaceable to ensure compliance with regulatory requirements, maintain ethical standards, and verify ac- curacy of information, automating various parts of this process offers significant benefits. AI-assisted contract writing can increase contracting speed, accuracy, and minimize costs. In 2024, I submitted an initial proposal to the Naval Innovation Science and Engineering (NISE) program titled Generative AI Application in Contracting and was invited to submit a full proposal to brief the command on my proposed project. The NISE program, authorized by public law, enables the director of the DoD Warfare Centers and Laboratories to utilize between 2 to 4 percent of all funds for the purpose of basic and applied research, technology transition, lab revitalization, and workforce development. The aim is to foster creativity, explore cutting-edge science and technol- ogy, provide proof of concepts, support high-risk research, and enhance the labs’ capabilities for future military and DoD missions. The problem statement explained how current contracting methods are limited in efficiency, adaptability, experience, and capability and lead to incomplete or contracting products of systemic inefficiencies that ultimately increase costs and risks for government stakeholders. The approach proposed utilizing Microsoft Azure, an AI/ML application that NAVFAC EXWC acquired to validate the performance of the platform by testing and evaluating, identifying best practices and current limitations, and presenting findings to the agency. The objective was to test and evaluate the use of Microsoft Azure to accelerate contract development and execution and the preparation of acquisition documents. The approach utilized an awarded contract through NAVFAC EXWC, and its customer, NAVFAC Information Technology Center (NITC) by obtaining DoD Impact Level 4/5 Microsoft Azure licenses, which allowed for a higher level of protection for controlled unclas- sified information (CUI) data. This innovative approach aligns with NAVFAC EXWC’s vision of accelerating innovation to enhance fleet effective- ness both at sea and ashore. By leveraging recently acquired technology, we aim to expedite innovation, delivering superior contracting products that enable faster project execution with reduced risks and costs. The approach would include taking Microsoft Azure training, requirement gathering/analysis, system design and architecture, data collection and processing, natural language processing and model development, generative AI implementation, template and clause development, integration with existing systems, and user interface design, testing, deployment and maintenance. The team would test and evaluate their prospective use-cases to determine if the Azure services could meet the project’s intent. To determine the success of these tools, the team would test their solution on previous successful documents and contracts to see if the AI tools could produce an elevated product. The tool suite would be used for advising on contract execution metrics using data from DoD applications. It leverages the Azure OpenAI service to research regulations, class deviations, and Government Accountability Office (GAO) cases. It would develop comprehensive risk decision trees to assist in determining acquisition strategies. Additionally, it uses “document intelligence” to train and test past contract documents, aiming to assess their ac- curacy and consistency. Finally, Azure’s OpenAI service assists in writing various contract execution documents, including market research, acquisition planning, justification and approvals, determinations and findings, and business clearance memorandums. The implementation of this project would present challenges through multiple fronts, including protecting data, fol- lowing laws and regulations, integrating with existing systems, timing concerns, and command investments. However, we thought of ways to mitigate them. To protect data privacy, we would utilize DoD impact level 4/5 licenses. Rules of engagement would be developed and workforce development training would be provided to prioritize compliance with laws and regulations and discourage the sole reliance of AI. Furthermore, due to a lack of integration with existing systems, and due to the period of performance of the contract ending in July 2025, we would need to test diligently and revise our project plans. As this project is a work in progress, the quantitative and qualitative aspects are not yet realized and have not been internally tested; however, according to initial studies conducted by the Deloitte Center for Government Insights, the amount of time saved varies significantly, ranging from 2.23 percent to 27.86 percent, depending on the extent to which governments invest in and allocate resources to artificial intelligence initiatives. Considering increased interest in developing the DoD’s AI capabilities, the level of investment would likely be on the higher end. With heightened AI readiness, the acquisition workforce would be better equipped to award contracts, allowing for the delivery of technical solutions quicker and improved resource allocation. NAVFAC EXWC uses standard procure- ment acquisition lead time (PALT) timeframes for each contract type that it executes. If AI was used, there would be a
May-June 2025 | DEFENSE ACQUISITION
53
Made with FlippingBook - Online Brochure Maker