31 In addition to these approaches, designating team-level ambassadors can help ensure that all stakeholders—and most importantly, the downstream users of the technology—are empowered and directly incorporated in the reporting process. Beyond sourcing new use-cases from the workforce and providing ongoing training, one core function of an ambassador and their respective team is to make critical long-term strategic decisions regarding the optimal point in time to invest in new Gen-AI tools. Given the rapid pace of Gen-AI’s advancement, technological obsolescence is a critical concern for enterprises making near-term investments. Understanding and monitoring market trends, including cost trends relative to the evolution of capabilities for specific vendors, is essential for determining the right moment to invest in a specific solution. Embed Versatility in the North Star Plan Last, these practices should be incorporated as early as possible into the enterprise’s North Star and updated throughout the implementation process as the integration roadmap and specific use-cases become more refined. At the beginning of the process, however, we recommend structuring Gen-AI versatility in an overall strategy by thinking through several areas of Gen-AI adoption, including: 1. Known Capabilities, which encompass applications of tools already utilized by peer organizations for similar tasks; 2. Theoretical Capabilities, in which the enterprise hypothesizes a series of use-cases that it believes Gen-AI tools can serve; 3. Unknown Capabilities, in which the enterprise anticipates discovering unforeseen use-cases over time that are either broadly applicable to the organization or unique to a specific team. More specifically, unknown capabilities pose a challenge to the military given the juxtaposition of Gen-AI’s nascency alongside the DOD’s operating culture, which rationally aims to mitigate risk given the high-stakes outcomes for many of its activities. The DOD’s acquisition system mandates that requirements are issued for a product before issuing contracts for its development and acquisition. However, for many Gen-AI tools, their capabilities follow a “jagged frontier” 35 pattern, meaning that use-cases are highly context-dependent. For any single process, therefore, capabilities are often difficult to determine until testing and evaluation are conducted, which can create additional complexities when considered within the DOD’s prioritization of requirements. The DOD, therefore, needs to identify operational vehicles to incubate Generative AI’s capabilities through experimentation. For example, one approach could involve tasking DIU or CDAO, while leveraging the existing AI infrastructure, to coordinate efforts across service branches or COCOMs in identifying pilot projects where promising use-cases may significantly expand the capability frontier for DOD Gen-AI tools, despite current uncertainty. Such efforts could help identify incremental operational improvements that grow each ‘edge’ on the jagged frontier,36 and also discover use- cases that result in an expansion of the entire frontier, resulting in innovation leaps that apply to multiple branches or COCOMs. Last, documenting and disseminating these discoveries in an organization’s North Star Plan, as well as a reporting system accessible to the DOD’s acquisition community, can help Program Managers and Contracting Officers more clearly scope requirements that align with the newfound capabilities. In return, this can foster competitive bidding processes for technologies that address breakthrough use-cases through the Commercial Solutions Openings (DFARS 212.70) process, which the Office of the Secretary of Defense recently identified as a paramount Software Acquisition Pathway (SWP). 37 Generative AI Adoption in the US Military