Forward deployed engineers: The crucial role in enterprise AI

Where do you want AI to create value in your organization? You can probably answer that quite easily: improve the customer experience (CX), reduce operational costs, increase productivity or find new sources of revenue.

But how do you turn those ambitions into AI initiatives that can be properly funded, closely governed, efficiently delivered and continuously measured?

That’s a much harder question to answer, and it’s where many enterprise AI programs are getting stuck.

A 2026 Gartner® article reports that “by the end of last year, at least 50% of generative AI projects were abandoned after proof of concept due to poor data quality, inadequate risk controls, escalating costs or unclear business value.”*

Often, value gets “lost in translation” because there’s a breakdown between what the business needs and how that can be delivered. Looking at the three typical layers of an AI program, strategy gets plenty of attention, as does technical delivery. But the biggest gap is often found between the two: translating business ambitions into well-defined, viable AI initiatives that delivery teams can put into production.

Forward deployed engineers (FDEs) can play a crucial role in connecting business intent with operational reality and technical execution to get AI projects into production successfully.

• ALSO READ: Making the case for AI: What your board and CFO need to know

Embedded with customers before or alongside major AI initiatives, FDEs help turn an AI ambition into something that can be built, deployed, adopted and measured — while preparing the organization for the change that comes with it.

It’s useful to distinguish the FDE from three roles already common in delivery organizations:

• A business analyst captures requirements, asking, “What do you need?” An FDE starts with the outcome: “What measurable business result are we trying to achieve, and what technical, operational and organizational changes will get us there?”
• A solution architect typically focuses on how to build the solution. An FDE asks an earlier set of questions: “What should we build, and why? How will we prove its value, and what will success look like?”
• A technical program manager coordinates the execution of defined work. An FDE helps determine what that work should be in the first place and keeps it anchored to the business outcome throughout delivery.

In short, FDEs bridge the commercial, operational and technical worlds of the customer and the delivery organization.

How the role has evolved

Palantir, the US software company, pioneered the FDE model in its early years, later formalizing it as a distinct engineering role.Since then, the model has spread to AI-native companies such as OpenAI and Anthropic, to hyperscaler partner programs at Amazon Web Services and Microsoft, and to consulting firms.

The basic idea hasn’t changed: Put an engineer close to the customer and the business problem they’re trying to solve. What has changed is the scope of that problem, and therefore the scope of the role.

Before GenAI, FDEs were typically customer-facing engineers, integration specialists and deployment experts rolled into one. Much of their job involved overcoming technical obstacles to get a defined product working in a defined environment.

With AI, however, the tough questions often surface much earlier. Which use cases are worth pursuing? How does the underlying business process actually work? Is the data good enough? How will value be demonstrated? And what will it take to move successfully from prototype to production?

Answering these questions requires a broader set of skills, and the role of the FDE has evolved from deploying software to delivering outcomes.

What the AI-era FDE does

The AI-era FDE is part engineer, part solution architect, part consultant, part product owner and part business translator. They’re accountable not only for making the technology work but also for helping decide what should be built, demonstrating why it’s worth building and creating a governed path to production.

Increasingly, that means bringing business leaders, employees, engineering teams and autonomous AI agents together around the same goal.

As AI agents become more autonomous, the role of the FDE becomes more important. Technical barriers are falling and experimentation is getting cheaper, making it easier to build something with AI. What’s harder is deciding what should be built, redesigning the processes around it, managing risk and governing the decisions. And increasingly, it means deciding which parts of the work belong to people and which to AI.

How an FDE finds the right AI opportunity

Every organization has made some version of this request: “We need AI to improve the CX.” That’s a reasonable ambition, but it gives an engineering team very little to build.

An FDE works backward from the business outcome to turn it into something that’s measurable and executable:

1. Define the outcome

Start by making “improve CX” measurable. That could mean reducing call-handling or onboarding time, improving your net promoter score or increasing first contact resolution.

2. Map the workflow

Understand how the work happens today: the customer journey, process flow, decision points, human bottlenecks and available data. This reveals where AI can make a real difference and where it probably won’t.

3. Identify the right interventions

Rather than applying AI everywhere, identify the specific interventions that are most likely to produce the desired outcome. These might include agent copilots, onboarding automation, claims triage, intelligent routing, document interpretation or knowledge retrieval.

4. Establish how you’ll prove value

Before you can justify a wider rollout or outcome-based pricing, you need a baseline and a way to measure what changes. How do you measure performance today? Which metrics are important? What data and instrumentation will you need, and what governance should be in place?

5. Turn the opportunity into executable work

By this point, “improve CX” has become something much more concrete: “Deploy an AI support assistant that reduces average handle time by 20%, increases first contact resolution by 15% and reports performance against agreed key performance indicators.”

Now the engineering team has something it can build, and the business knows what success will look like.

In practice: Building the operating model

Putting this into practice requires a few changes to the AI operating model:

• Create a formal translation layer. Most organizations have strategy and engineering teams, but far fewer have FDEs who act as AI translators or outcome architects responsible for connecting the two.
• Standardize the path from idea to production. Use a repeatable sequence — assess, design, build, manage and optimize — rather than reinventing the delivery process for every new opportunity.
• Put governance at the front of the process. Before execution begins, AI initiatives should pass through qualification, architecture validation, and delivery readiness and risk assessments, with the intended business outcomes clearly defined.
• Develop hybrid talent. FDEs combine engineering credibility and AI platform expertise with consulting skills, commercial awareness, industry knowledge and the ability to bring different stakeholders together.

NTT DATA’s Agentic AI Studio embodies this approach. It provides a secure, end-to-end platform and operating model for designing, building, deploying and scaling AI systems with built-in governance and risk management.

Don’t just build an AI practice. Build an AI operating model.

My advice to leaders building AI transformation teams today is this: As AI technology becomes more widely available, just having access to it is no longer a differentiator. What counts is your ability to find the right problems, translate ambition into executable work, measure the results and repeat what works.

That also changes how you spend on AI. Instead of funding a succession of disconnected AI projects, invest in the capability to repeatedly turn promising ideas into measurable outcomes. A small team of FDEs embedded in priority areas of your business can identify valuable opportunities, map the path to delivery and establish patterns that can be reused elsewhere.

• ALSO READ: The real AI opportunity: Reinvent how your organization operates

What to do next

Get in touch with us to enable the role of forward deployed engineer in your organization.