Forward deployed engineers (FDEs) have become one of the most debated topics in enterprise technology, driven in large part by the visibility of Palantir’s model. As generative AI and agentic AI reshape enterprise architectures, many CIOs and CTOs are asking whether this approach represents the future of software delivery. In fact, the reality is more nuanced. Forward deployed engineers are highly effective in the right context, but potentially risky in the wrong one. Understanding where they fit requires a clearer view of how AI is actually being used inside enterprises today.

What Forward Deployed Engineers Really Do

At its core, the concept of using forward deployed engineers is straightforward. Forward deployed engineers are software engineers embedded directly within a client’s business operations. They work alongside business teams and make real-time adjustments to the code and platforms they support. This allows systems to continuously adapt to changing business needs, rather than waiting for formal release cycles.

This model has been proven in environments such as Palantir’s, where the tight coupling between software and operations demands constant iteration. In these contexts, forward deployed engineers are not just helpful, they are essential. They ensure that the platform evolves in lockstep with the business.

The controversy arises when organizations attempt to generalize this model across all enterprise systems.

The Two Mandates Driving AI Adoption

To understand where forward deployed engineers make sense, we need to start with how CIOs are approaching AI. In most enterprises, there are two distinct mandates.

The first is optimization. CIOs are using AI to do what they already do, but more cheaply and effectively. This includes software development, application management, and infrastructure operations. Code generation tools, automation platforms, and AI-enabled interfaces are being layered into existing systems to improve productivity.

In this mode, the underlying tech stack remains largely unchanged. AI acts as an enhancement, rather than a transformation. As I have noted previously, much of today’s adoption falls into this category, where AI serves as a feature extension, rather than a fundamental shift in operations.

The second mandate is more transformative. It involves building what I describe as “agentic native” environments, where AI is not just assisting processes, but actively driving them.

Inside the Agentic Native Environment

Agentic native environments are structurally different from traditional enterprise systems. They typically begin with an ontology, which functions as a private large language model that understands not just the data of the organization, but its meaning. This creates a level of observability that goes far beyond what enterprises have historically achieved.

We often see paired with the ontology a digital twin, a virtual model of the organization’s processes or infrastructure. This enables simulation, allowing companies to test changes and anticipate outcomes before implementing them in the real world.

On top of this foundation sits a fabric of agents. These agents act on the insights generated by the ontology and digital twin. They are highly dynamic, created and destroyed as needed to execute specific tasks.

The result is an environment that is continuously evolving. The ontology expands as new data and knowledge are incorporated. The digital twin becomes more sophisticated over time. Agents are constantly being generated to respond to new conditions.

This is not a static system; instead, it is a living one.

Why Forward Deployed Engineers are Essential in the Agentic Native Context

In such a dynamic environment, traditional governance models break down. You cannot rely on slow, controlled release cycles when the system itself is changing in real time.

This is where forward deployed engineers play a critical role. They provide the human layer of oversight and control. Because they are embedded in the business, they understand both the operational context and the technical architecture.

Their role is to guide the system, adjust it as conditions change, and ensure that the actions taken by agents align with business objectives. They also help manage the continuous evolution of the ontology and digital twin, ensuring that new data and insights are correctly interpreted and applied.

Without this level of embedded expertise, organizations risk losing control of a system that is inherently dynamic.

Mistaken Use of FDEs in Traditional Systems

The mistake many organizations make is assuming that forward deployed engineers are broadly applicable across their entire technology estate.

Traditional systems are designed for stability. They evolve slowly, using configuration, APIs, and controlled updates, rather than real-time code changes. Governance models are built around rigorous testing, validation, and release management.

Introducing forward deployed engineers into this environment can bypass these safeguards. Real-time changes to production code create security risks and the potential for significant failures. These systems are not designed to absorb that level of dynamism.

As a result, forward deployed engineers are not only unnecessary in traditional stacks, but they can be actively harmful.

A Signal of a Broader Shift

It is important to view forward deployed engineers as a signal, rather than a solution. Their presence indicates that an organization is moving toward an agentic native operating model.

This shift requires more than just embedding engineers in the business. It involves rethinking governance, investing in simulation capabilities such as digital twins, and building systems that can safely support continuous change. Forward deployed engineers are one component of this model, but they are not a substitute for it.

The industry discussion around forward deployed engineers has been valuable, but it often misses the central point. The question is not whether they are good or bad; the question is where they fit.

They make sense in environments where systems are dynamic, continuously evolving, and deeply integrated with business operations. They do not make sense in traditional systems designed for stability and control.

CIOs and CTOs should resist the temptation to adopt this model wholesale. Instead, they should align their talent strategy with the nature of their technology environment.

In the age of agentic AI, precision in operating models will matter far more than following the latest trend.