Article

Forward Deployed Engineering (FDE): The Missing Layer in Your AI Operating Model

A Forward Deployed Engineer (FDE) is the embedded technical engineer that activates and operationalizes AI capabilities—LLMs, Agentic AI, and vibe coding—connecting them to business systems, data, and workflows. They are the ones implementing AI governance across the key components of Strategy, Design, Build, Integrate, Go-live, and Run.

Enterprise leaders are desperately trying to wrap their heads around why AI isn't paying off — and how to claw their way out of what feels like pilot purgatory. They are not alone: fifty-six percent of CEOs say their companies aren't yet seeing a financial return from AI investments. Microsoft’s Frontier Firm Initiative puts it plainly: pilot-rich but transformation-poor.

HFS leaders Phil Fersht and Saurabh Gupta offer some solace. Their recent article, Stop treating FDE as optional: Your AI Flywheel will not spin without it, names the missing layer most AI operating models lack: Forward Deployed Engineers (FDEs).

The Claim

"93% of enterprises are stuck in AI pilot purgatory. The missing layer is not better models or bigger budgets. It is Forward Deployed Engineering, and the firms that crack it at scale will own the recurring revenue layer of enterprise AI."

Phil Fersht & Saurabh Gupta

HFS Research

FDEs are the embedded technical engineers that activate and operationalize AI capabilities—LLMs, Agentic AI, and vibe coding—connecting them to business systems, data, and workflows. They are the ones implementing AI governance across Strategy, Operating Model Design, Build, Integrate, and Run. And for businesses who want to scale AI, FDEs are non-optional.

Without them, even the most sophisticated AI capabilities stall at the threshold of real business impact. LLMs can reason and generate code but cannot connect themselves to governed data or act in regulated industries without human intervention. Agentic AI can orchestrate across systems but accumulates risk around decision rights and accountability without proper governance design. Vibe coding can produce working agents at speed but creates fragmentation, compliance exposure, and technical debt without standards and guardrails. FDEs are what close all three gaps.

Our Verdict

The HFS core argument lands: the gap in AI transformation is not technical; it’s operational. We agree, but we would go one step further: most importantly, it is organizational.

Business Process Outsourcing (BPO) is out, "Expertise Dense" is in.

The traditional BPO model — decades of people-intensive service delivery built on sheer headcount — is being replaced, and fast. HFS calls the alternative 'Expertise Dense': rather than throwing bodies at a problem, you deploy compact, highly AI skilled teams. We agree. It delivers better outcomes, faster, at lower cost.

The FDE unicorn problem.

The HFS ideal FDE profile, who is fluent in LLMs, agentic platforms, and vibe coding, with genuine operational and industry depth, is real but rare. Organizations need Forward Deployment Teams (FDTs): multidisciplinary squads that collectively cover the skill set.

The design question nobody is asking.

HFS positions the FDE as the site architect — but who designed the building? Most organizations lack that capability in their current transformation teams. They define what they want the process to do based on how it has always worked, then ask the implementation team to make AI fit.

The change management challenge is still underestimated.

The FDE model by HFS describes a world where the business users, process operators, and managers are passive recipients of a transformation. Bain's recent analysis, Want More Out of Your AI Investments? Think People First, makes the case that organizations treating AI adoption as a people and readiness problem outperform those treating it as a technology deployment.

What To Do Now

HFS's call to action is the right instinct: treat FDE as a strategic capability, not an implementation afterthought. Here is how to act on it.

Redesign before you deploy.

Audit the process before you automate it. If the process is broken, complex, or built around constraints that no longer apply, fix it first.

Use FDTs while you develop your cadre of multi-skilled FDEs.

Assess your current skills inventory across LLMs, agentic AI, vibe coding, and operational depth. Identify gaps. Build multidisciplinary teams to cover them now, while developing a longer-term plan to grow the next generation of true FDEs.

Match the reimagination approach to the magnitude of the opportunity.

Not every process needs a white-sheet redesign. Some benefit from staged AI renovation. Others can be rebuilt from SOPs using agents.

Invest in people readiness alongside technical deployment.

Structure change management as a parallel workstream, not a post-launch activity. Engage business users early in the process planning. Organizations that get sustainable returns on AI are the ones where the people responsible for running the system understand it well enough to improve it.