Contract Assistant GEnerative AI (CAGE-AI) Tool

Mile Two’s generative AI contract assistant (CAGE-AI) proposal delivers an innovative, first-of-its-kind prototype that will provide pre-solicitation workflow assistance by helping the analyst perform market research and develop artifacts in support of a solicitation via an intuitive user interface powered by generative AI and natural language processing. The results will improve the pre-solicitation process through increased efficiency and consistency.The Air Force Research Laboratory is grappling with the challenge of outdated acquisition and contracting methods, which are heavily reliant on manual processes. In a recent report, the U.S. Government Accountability Office found that DoD weapon systems “routinely take much longer to field, cost more to buy, and provide less capability than initially intended” (GAO, 2021, p. 3). This leads to an inability to meet the rapid needs of modern defense requirements. A study from RAND also found a heavy focus of acquisition reform is on limiting program cost overruns and schedule delays (Wong et al., 2022). Pre-solicitation activities such as market research and requirements definition, are critical for an efficient and effective acquisition process (GAO, 2021). Effective pre-solicitation planning reduces the likelihood of cost overruns and schedule delays, increases overall acquisition efficiency, and ensures better overall alignment. Further, “While the Defense Department’s acquisition budgets have grown significantly in recent years, there are not enough contracting officers to manage their ever-increasing workload” (Obis, 2024).M2 proposes the Contract Assistant GEnerative AI (CAGE-AI) Tool, a purpose-built application designed to make acquisition data highly discoverable and streamline the pre-solicitation process. The vetting process requires gaining an understanding of what currently exists in the contract landscape. This process is inefficient, insufficient, and time consuming especially when performed by non-contracting professionals such as researchers in the lab, personnel in operational settings, and programmatic support.CAGE-AI addresses this information retrieval problem by providing a vector database with large language model analytical capabilities to organize and explore acquisition data. Contracts and their associated metadata will be translated into highly discoverable segments which will allow CAGE-AI to read the contents of contract documents and allow users to run advanced queries against them. Once an idea has been established and vetted, the same group of people will be responsible for developing the requirements documents and solicitations that follow. Streamlining this part of the process with a flexible framework for creating, capturing, and standardizing pre-solicitation data will improve efficiencies, enhance collaboration, and reduce clumsy handoffs between problem holders and the contracting team.