A New Way to Make AI Actually Work in the Real World

• By Mohamad Ali, Senior Vice President and Head of IBM Consulting

• IBM is introducing a delivery model built for the AI era: small, senior teams that rapidly turn strategy into results through hands-on execution

IBM Consulting and Pearson team

Enterprise AI is at a tipping point. The investment is massive and experimentation is everywhere but deploying quickly remains a challenge. The issue is not the vision nor the technology. It is the operating model.

For decades, scale in delivery came from labor. Add more people, get more output. Every commercial model was built on that logic. AI changes the equation. Output now depends on an organization's operating model: how well teams build and coordinate agents, enforce governance, and turn raw capability into measurable business results. Most enterprise delivery models are still built for the labor era.

But the industry is chasing a job title: "forward deployed engineer (FDE)." IBM has always had FDEs, including Fellows and Distinguished Engineers, embedded directly in client work, and FDEs extend that practice into a repeatable, scaled model. Now, IBM Consulting is launching a new approach for how AI gets delivered called Forward Deployed Units (FDUs). An FDU isn't a person; it's a pod. Humans at the edges with a digital workforce of specialized agents in the middle, handling coding, evaluation, testing and documentation under human direction.

Human and Digital working as a team — by design

The composition is the point. It lets a six-person pod do the work of a 30-person team at materially better economics, with methods that sharpen with every engagement. It's how AI becomes a scaling factor, not just an assistant.

FDUs are already at work with Riyadh Air, Nestlé, Heineken and Pearson, moving AI from isolated pilots into production at scale. And we are now deploying them at global scale, from Asia Pacific to Europe to the United States, and rapidly increasing the number of FDUs we are putting into the field.

The rise of the "forward deployed engineer" and why it's not enough

Recently, there has been a rapid rise of the FDE. At its best, this role blends engineering, consulting and business expertise into one. It's someone who can understand a problem, design a solution, and build it directly in the environment where it will be used in real time. Enterprises are no longer satisfied with strategy alone and need people who can take AI into production. But focusing on a single role misses the broader issue. The rise of the FDE is not the solution, it is the signal. It shows that the way technology is delivered must change.

The underlying issue is not about talent; it's systematic. No individual can solve for fragmented data, complex architectures, governance requirements and the need to move from idea to production in days instead of months. What's needed is a delivery model that connects strategy, engineering and business context into a single system. That's the unlock businesses need and what IBM's FDUs are designed to do.

Each team of FDUs is accountable for real business outcomes. They combine business domain specialists who rethink processes, architects who connect strategy to execution, and engineers who build and scale solutions. By working alongside client teams as an extension of the organization, they change how work gets done. And at their core, they combine people, platforms and AI agents into a unified system, where human expertise shapes the work and AI accelerates its execution.

To support this, IBM maintains a dedicated technical career track for FDUs and recruits from top global engineering and technical universities to ensure a pipeline of the highest-caliber forward‑deployed talent.

From projects to continuous execution

Traditional models separate thinking from doing. Strategy is handed off, and context can be lost. FDUs collapse that model. The same team designs and builds solutions, and progress is measured in working systems rather than deliverables, which is crucial for constantly evolving agentic AI systems. They require ongoing tuning, governance and integration into live workflows, so delivering them is not a one-time project, but continuous execution.

This is where many approaches can fall short. FDEs can help get systems up and running, but AI does not stop at go-live. When delivery relies on individual roles, there is often pressure to move on after a system has been launched. This creates a gap between implementation and sustained performance.

FDUs are designed to address that disparity. They bring solution development, ongoing operation and client capability into one model that sustains value over time, not only at launch. And because client teams work side-by-side with senior practitioners throughout the engagement—not through handoffs or post‑delivery knowledge transfers—they build lasting internal capability to operate, evolve and scale AI long after the FDU has left. This turns every engagement into both an execution engine and a transformation accelerator.

Why platform matters as much as talent

Embedding teams is necessary, but not sufficient. To scale AI, teams need a shared foundation that offers speed, consistency and governance.

At IBM, FDUs run on IBM Consulting Advantage, an AI-powered delivery platform that provides reusable assets, AI agents and industry accelerators. It enables faster delivery and repeatability, turning isolated wins into enterprise-wide value, ensuring that teams don't start from scratch but build on a common, scalable system.

This combination of senior consultants, FDE‑level technical talent and IBM Consulting Advantage is something only IBM brings together in one model.

The next chapter of AI will be defined by execution

The next phase of AI won't be defined by models alone; it will be defined by the ability to turn them into sustained business value. The conversation is already shifting beyond tools and talent toward delivery systems.

FDEs are part of that story, but they aren't the whole story. Organizations that lead will bring together the right teams, platforms and operating model into a single system.

That is what we are building with FDUs. It's how we move from experimentation to execution, and how we help clients turn AI ambition into real, measurable outcomes.