37 The DOD’s contracting efforts, adoption areas, and timelines can guide the private sector on where and how to integrate Gen-AI tools. Moreover, understanding the major challenges that the DOD faces in adopting these tools offers additional unique insight. Below we discuss these challenges in the context of the private sector and outline several takeaways for companies seeking to adopt the technology for enterprise functions similar to those that drive the DOD. Thoroughly Evaluating the Tech Stack Early is Crucial Companies seeking to develop a thorough understanding of where and how to prioritize integration of Gen-AI tools must thoroughly understand which areas of the organization are best suited for them. This helps establish the cost, timeline, scope, and most importantly potential impact of the integration. It also helps businesses identify any enabling technologies and services needed to coordinate alongside a Gen-AI integration, so that coordination can occur across vendors to optimize the effectiveness of any implementation. Most important to consider is how the scope and complexity of a tech stack assessment may change depending on the size and maturity of the organization. Large-cap and older enterprises, for example, often have more enterprise resource planning tools and disparate data repositories than smaller organizations or startups. For reasoning models, disconnected and outdated data silos make it harder for implementation teams to identify and capture critical data pools within the enterprise, and then integrate them to fine-tune the model. RAG applications can help with this through retrieving isolated datasets and chunking them into a language map, but it may struggle to gain access to these detached data silos within the enterprise if significant barriers exist to navigating the company’s broader digital files. For AI Agents, navigating a much larger web of software programs, such as CRMs, file sharing services, and cloud infrastructure, can drastically increase the number of potential failure-points in a self-prompting process, leading to sharp increases in error margin. However, despite facing a more challenging integration environment, large incumbents have the advantage of existing resources such as technical labor and finances required to thoroughly plan and invest in modernizing their tech stack and better position the enterprise for Gen-AI integration. These firms can even develop special-purpose Gen-AI tools of their own where organic development makes sense, as Dimitri Alves, General Manager for L3Harris’ Microelectronics division remarks: On the other hand, while small-to-mid-cap enterprises lack the resources to develop many of the same organic capabilities, they have the advantage of more streamlined resources, processes, and tools, and therefore do not face the same entrenched complexities as their larger counterparts. says Tony Morash, Director of Business Development and Strategy at Aeronix Technologies Group’s C6ISR Division, who brings experience working in the intelligence community along with multiple Fortune 500 aerospace and defense contractors. He adds: We have a few home-grown tools, one of which is particularly helpful for our engineer workforce at the outset of projects. It supports them by offering advice for keeping within the project guidelines or suggesting best practices and key considerations in kicking off projects. One advantage we have for LLMs, is that this is the cleanest data environment I've ever experienced. “ “ Generative AI Adoption in the US Military