en.Wedoany.com Reported - As enterprises accelerate the deployment of autonomous artificial intelligence systems, large language model (LLM) providers are embedding AI experts into client teams through Field Deployment Engineer (FDE) services to help create, customize, and launch AI offerings. Microsoft and Amazon Web Services (AWS) have recently announced large-scale investments to drive this service model forward.

This week, Microsoft launched a $2.5 billion venture—Microsoft Frontier Company—which will deploy 6,000 experts to collaborate with clients in designing, innovating, deploying, and continuously improving AI systems based on their business objectives. The service focuses on "frontier transformation," helping clients build intelligent platforms using proprietary data, internal expertise, workflows, and decision-making processes. Judson Althoff, CEO of Microsoft's commercial business, stated in a blog post that based on FinOps principles, the service helps users observe, govern, manage, and protect AI tools across their entire technology stack, with intelligence compounding over time. The platform is "model-diverse, open, and heterogeneous," allowing clients to choose from ChatGPT, Claude, Microsoft Copilot, or other open-source or industry-specific models without being locked in. Additionally, client data and intellectual property are protected and will not be used to train Microsoft models. Microsoft will leverage partnerships with Accenture, Capgemini, EY, KPMG, PwC, and others to scale the platform. Althoff noted that early adopters such as the London Stock Exchange Group (LSEG), Land O'Lakes, Unilever, and Novo Nordisk have already achieved measurable results. For example, AI embedded in LSEG Workspace helps financial experts pose complex questions and quickly obtain answers based on structured and unstructured financial data, with the underlying foundation iteratively optimized through client feedback and real-time user testing, accelerating each cycle and improving model quality and scope.

Meanwhile, AWS announced a $1 billion investment in its new AWS FDE platform. Francessca Vasquez, Vice President of Frontier AI Engineering and Services at AWS, explained in a blog post that unlike traditional consulting, which treats each deployment as an independent project, AWS FDE focuses on long-term construction, helping clients transition from "observers" to "co-builders" and then to "autonomous operators," achieving "AI self-sufficiency." The platform is agent-first, aiming to compress timelines from months to days, with derived business intelligence compounding to support future projects. Embedded engineers—many of whom built AWS AI services—will validate and guide projects. Clients can access runbooks and architecture documentation, with a semantic layer connecting to their data sources to create knowledge graphs that AI agents can reason over. Vasquez emphasized that domain expertise resides in clients' code, agents, and systems, preventing institutional knowledge loss when employees leave. Security tools provide hardware-based isolation and end-to-end encryption. The service is not designed for organizations experimenting with AI, but for those that have moved past the experimental phase and need production-grade AI systems to run real business processes.

Thomas Randall, Research Director at Info-Tech Research Group, noted that the gap between AI investment and return on investment is widening, putting pressure on organizations to demonstrate the production value of AI deployments. In this context, FDEs from vendors like Microsoft and AWS will leverage their deep product knowledge to compress learning curves, establish reusable processes, and build transferable capabilities. He also stated that Info-Tech research shows 77% of organizations lack enterprise-level AI strategies, and FDEs will address this by focusing on clients' running AI systems, reference architectures, runbooks, and other deliverables.

For system integrators (SIs), technology analyst Carmi Levy pointed out that SIs have maintained high-margin relationships with clients for decades, and it "makes a lot of sense" for hyperscalers to try to capture some of that business for themselves. These vendors are actively seeking ways to strengthen client lock-in and create more opportunities to embed themselves in client operations and decision-making mechanisms. Randall said that the value SIs provide lies in broader cross-system integration knowledge, change management, and project scaling, with deliverables that are more strategic and broader in scope. While there is overlap, Microsoft will work closely with global SI partners. Investment gaps and implementation complexity mean hyperscalers need to offer more white-glove services to drive client adoption.

Levy advised that for clients who have already decided on a specific AI technology stack and are willing to go the single-vendor route, these platforms may be worth considering, but using them could come at the cost of reduced potential choices, limiting long-term options. He suggested IT decision-makers deeply understand the agent delivery capabilities of Microsoft and Amazon compared to SIs, and investigate whether the underlying motivations truly align with clients' best interests. Randall recommended that enterprises consider desired outcomes: FDE options are best suited for organizations looking to quickly and effectively build products, while SIs are needed when these organizations need to scale within complex enterprise processes. Additionally, FDEs are not suitable for organizations still studying fundamental AI strategy questions or wishing to maintain cloud neutrality.