Today’s conversation about Forward Deployed Engineers (FDEs) is part of an answer, but it is not the whole answer.

Published 18 June 2026

For thirty years, labor was the scaling factor for delivery. More humans, more output. Every commercial primitive in this market—time and materials, t-shirt sizing, pod month, fixed price per full-time employee (FTE)—was calibrated to that fact.

AI is now the scaling factor. Output is no longer linear in human capacity; teams shape it by orchestrating agents, enforcing evaluation discipline, running reproducible methods and converting raw capability into measurable business outcomes. Most enterprise delivery models—IBM’s included—are still calibrated to the old model.

The teams that rebuild around the new scaling factor capture the AI dividend. The teams that don’t leak it through reduced hours billed, with no value capture on either side of the contract, lose it entirely.

That gap is real. Today’s conversation about forward-deployed engineers (FDEs) is part of an answer, but it is not the whole answer. Naming a new role does not shift the dividend. Rebuilding the unit of delivery around the new scaling factor does.

The shift happening across the industry is fundamental. We’ve moved from assisted development (code completion, prompt engineering) to asset-driven delivery (reusable components and accelerators) to agentic capabilities (autonomous systems that reason, plan and execute).

The next wave is emerging. Spec-driven development and context engineering will span the entire software development lifecycle. Agents won’t just help us code, they’ll help us design, test, govern and operate entire systems. The intersection of these “agent skills” and human guidance is where the next evolution of the technical expert needs to reside.

The forward-deployed engineer has been called the “hottest job in technology.” Yet even as enterprises wake up to the need for strategy and execution to coincide, a single role won’t meaningfully shift the outcome.

After analyzing the market and how we’ve delivered technology and value to thousands of enterprises, we see the gap clearly. Implementing and adopting technology at scale and pace doesn’t come down to one role. Instead, it requires a strategic cohort of forward-deployed roles. We call them forward-deployed units (FDUs).

FDUs are the bridge between AI platforms and enterprise transformation. The pattern is not simply “more engineers on site.” It is a shift from advisory-led transformation to engineering-led transformation, where small teams are embedded in the client context and paired with platforms, accelerators, change management, industry workflow knowledge and deployment capacity.

Each role is designed to avoid the pitfalls of the old delivery models, ensuring enterprises build scalable capabilities that deliver the value the business needs.

The first is the client outcome lead. These leaders are senior industry or functional domain experts. They own the business outcome that the FDU is engaged to deliver—the close-cycle compression, the customer-acquisition cost reduction, the throughput uplift and the regulatory exposure closed. They lead the squad day-to-day and are accountable in the client’s quarterly business review (QBR) for the result, not just the work.

The client outcome lead closes the accountability gap. Outcome-based contracts work when someone on the squad owns the outcome end-to-end. Without that role, the engagement defaults to delivering the work and hoping the outcome follows. The client outcome lead turns the outcome from a contract clause into a person.

The second is the business domain specialist. They have a deep understanding of the industry and process domain. They are senior specialists who bring real industry and functional domain expertise, not generic process consulting capability.

Business domain specialists solve the reimagined process challenge. These specialists collaborate with and challenge your process owners to rethink objectives and outcomes.

The third is the forward-deployed architect. They aren’t traditional enterprise architects who draw diagrams and disappear. Forward-deployed architects are senior technical leaders who:

  • Work embedded in your environment to shape AI and platform adoption.
  • Bring deep engineering authority—they know what can be built across your technology investments (including multiple technology providers).
  • Connect business objectives directly to technical implementation.
  • Design solutions with proven assets and patterns, not theoretical approaches.

Forward-deployed architects solve the strategy-to-execution gap. When the person designing your AI architecture has the business domain context, engineering depth and access to proven, reusable assets, you avoid strategies that look good but can’t be implemented.

The fourth role is the forward-deployed engineer (FDE). The presence of the forward-deployed architect sharpens and accelerates the work of FDEs, who then:

  • Orchestrate AI assets, agents and applications into cohesive business solutions.
  • Implement with agentic development approaches—leveraging AI to accelerate AI implementation.
  • Ensure that solutions align with architectural plans and scale effectively, considering technologies from multiple technology providers.
  • Embed with your teams to build capability that meets specific business use cases, not just functional deliverables.

This role solves the implementation velocity problem. These engineers aren’t junior developers augmenting your staff. They are senior engineers who can navigate complex enterprise environments and make autonomous technical decisions aligned to business value.

Alongside these forward-deployed roles, specialized data designers and applied AI specialists focus on preparing enterprise data for agentic systems. They also build AI-powered applications with agentic workflows.

These roles form the cast. The architecture is what makes them an FDU.

Inside an FDU, humans live at the edges of the workflow. The business domain specialist and the forward-deployed architect frame the work, curate the context and define what good looks like. The forward-deployed engineer and the data designer ensure what is produced is correct, governable and ready for production. The middle of the workflow—the work itself—is done by agents executing on codified methods against a curated context.

The client outcome lead sits above this workflow and is accountable for the result it is meant to produce. The squad delivers the work and the outcome lead owns the result and is in front of the client at the quarterly review.

This model is not a “human in the loop” bolt-on. Junior throughput is no longer the unit of capacity. Senior judgement at the edges is. That structural choice is the source of the dividend.

The other change is that the pod is the standing operational unit of delivery. Not a project construct that disbands at engagement end. Pod composition flexes with the engagement; pod identity persists across them. That persistence is what allows the knowledge layer to accumulate in the same hands across multiple engagements rather than being rebuilt every time. This operating model is what makes the second value stream operationally real rather than aspirational.

These FDUs are powered by IBM Consulting® Advantage (ICA)—the AI-powered platform that equips IBM consultants with industry, role and business-domain-specific AI assistants, agents and applications. It includes capabilities such as context engineering, process engineering and agent studios.

Agents work inside the same security perimeter and access controls as your own people—no special privileges, no exceptions. The platform plugs into your existing security and data governance, so your chief information security officer (CISO) doesn’t need to approve a new control plane.

In the FDU model, ICA becomes more than a productivity tool. It becomes the common platform for reusable methods, agentic application builds, context and memory engineering, governance, observability, marketplace applications and repeatable offering assets. When deployed in a client environment for ongoing reuse, ICA becomes an enterprise advantage. This service allows IBM Consulting Advantage to be placed under the client’s banner as the operational vehicle for the assets we license back to them.

This full-stack team is what modern AI implementation requires—not just consultants or technical talent, but integrated capability delivery.

As these FDUs set the benchmark for how modern engineering teams adapt to rapid industry shifts, IBM Consulting is investing in building these squads at scale. The investment draws on internal expertise, external talent acquisition and next-generation development. This approach recruits top STEM graduates from leading universities through an applied AI associate program.

The program is intentionally designed to give new graduates a clear and fast-track career pathway toward the forward-deployed roles. It also provides focused enablement and hands-on immersive apprenticeships within FDUs operating inside client environments.

The client outcome lead role draws from a different talent pipeline. Senior industry leaders, often coming from former client-side executive positions or from IBM Consulting’s senior partner ranks, bring the credibility required for outcome ownership. They lead the outcome conversation in the client’s quarterly business review. They are not graduate hires; they are seasoned outcome owners, recruited and developed through a separate succession plan.

FDUs are not generic in their capabilities. They are configured to serve industry and functional-domain business outcomes:

  • Finance transformation FDUs focus on forecasting, financial close, journal automation and finance data orchestration.
  • Customer transformation FDUs focus on Salesforce, CRM journeys, Einstein, MuleSoft, Slack and customer or employee experience agents.
  • IT operations FDUs focus on AIOps, runbooks, observability, autonomous remediation and site reliability engineering (SRE) workflows.
  • Core banking modernization FDUs focus on mainframe integration, core data activation, compliance, payments, deposits, lending and API and event modernization.

Every FDU engagement leaves the client with two things, not one. The first is the business outcome that you paid for—features shipped, processes automated, systems running. The second is a reusable asset kit that stays with you: the captured business context, the tests that prove the AI works, the agent configurations, the methodology refinements. The first is what you measured. The second is what makes every future engagement faster, cheaper and less risky.

Most other consulting models leave only the first.

We do not just sell this model. We run it ourselves. IBM Consulting is Client Zero for the AI-augmented operating model. We use our own FDU pods, our own agent fabric, our own context layer and our own commercial wrappers, used alongside the technology partners and ecosystem components any client would use. The same value test that we ask clients to apply is one we have already applied to ourselves.

IBM publicly reported more than USD 4.5 billion per year in productivity gains, captured internally. This architecture spans our own delivery operations with the same architecture that we are proposing to clients. This reality is not a forecast or a model. It is reflected in the financials that we have already reported.

When we walk into a client conversation, we are not selling slides—we are reporting on results that are already on the public record. The architecture, commercial wrappers, reward structure and IBM Consulting Advantage platform are all stress-tested inside our own delivery before they appear in a client engagement. That alignment and public proof is the difference.

The productivity captured internally also fuels a flywheel. Margin freed up by AI-augmented delivery is reinvested into the platform—deeper context bundles and more codified methods—and into reskilling, as the bench moves up the pyramid. This reinvestment also strengthens industry depth, with more verticals solved once and deployed many times.

Each turn makes the next FDU more productive than the last. That compounding asset layer—running first on our own books, then with you—is what turns a one-time dividend into a durable advantage.

When the same team that designs your business solution also engineers it using proven assets, you eliminate the translation gaps that prevent AI initiatives from succeeding.

Forward-deployed architects and engineers are senior-to-principal level technical leaders who have business domain context, can make complex architectural decisions, navigate organizational politics and drive through ambiguity.

These teams use agentic tools to build agentic solutions. They practice what they’re implementing for you. This approach delivers faster execution, higher quality and teams that understand the operational realities of the technology they’re deploying.

Tech and model vendors are paid when you consume more. Model providers charge for tokens, hyperscalers charge for compute and tool vendors charge for seats. Their business models are calibrated to your consumption.

Ours is calibrated to your outcome. IBM Consulting is paid when your AI dividend lands in your P&L. Outcome-coupled commercial wrappers, pod composition that shifts to capture productivity (we don’t bill for capacity we no longer need) and an asset layer that accumulates in your domain as your property. Aligned, not extractive. No other AI transformation partner can credibly claim this advantage.

FDUs are not billed by the hour. You pay for outcomes—features delivered, throughput sustained at a target, ongoing operation of a defined system or licensing of the reusable assets. These assets deploy in your environment through IBM Enterprise Advantage. The pricing matches the value that you capture, not the time we invest.

In any single engagement, several of these wrappers are typically used in combination. Per story for fine-grained delivery, throughput service-level-objective (SLO) for sustained product flow, IP subscription where the client wants to reuse the assets across other parts of their organization. Composition is the unlock. No wrapper works alone. It is how value capture is preserved as underlying productivity normalizes and per-unit pricing falls.

The alignment is not just contractual. It runs through how our teams are measured. The traditional consulting metric—resource usage, hours billed against capacity—actively penalizes a team that uses AI to compress effort. Their resource usage drops and so does their bonus. So we changed it.

Teams working on outcome-based engagements are measured on profitability and on the IP that they generate. The patterns, evaluation suites and agent configurations that they create are added to our central library and reused across the unit. Capacity usage is no longer the dominant metric. The same team sitting inside your engagement has its compensation calibrated to the outcome that you want.

The client outcome lead’s compensation is the most outcome-weighted of any role on the squad. Concentrated on the business result delivered, the asset bundle handed over and client satisfaction with the outcome reached, not on activity metrics. The role exists to own the result with a compensation structure that makes ownership real.

Your teams learn by working alongside experts, not by receiving knowledge-transfer presentations after the fact. You’re building an internal capability to sustain and evolve AI solutions after the engagement.

As the industry moves toward agentic AI at enterprise scale, the question isn’t whether to adopt these technologies or which singular role is “hot.” It’s choosing partners that have evolved their own delivery models and think AI-native and agent-native for your business use cases. These partners also have the engineering expertise to execute in a complex data and technical environment that isn’t just their own stack.

There is a second question, less obvious but just as critical as you progress. As you adopt the FDU model and the AI dividend starts landing in your P&L, the bottleneck shifts. It moves from capacity (which the FDU solves) to three places that have always existed, but the capacity problem kept out of view.

  • Deciding what to build. The cost of building the wrong thing now outweighs the cost of capacity. The financial weight shifts upstream from delivery to portfolio governance. Building the wrong thing faster is now the most expensive mistake.
  • Architectural integrity. AI writes more code, faster, across more teams and partners, but someone still owns the digital estate. Architectural drift, AI governance, model failure, regulatory exposure—these problems belong to authority, not delivery.
  • Organizational uptake. The rest of your business cannot consume change at the pace IT can now produce it. Faster delivery into a slow-adopting organization does not move the dividend.

The FDU model addresses the capacity bottleneck. The partnership we propose addresses the next three problems. This model is what an evolved delivery partner looks like in this era. One that has solved the bottleneck you are working through today and is already thinking about the bottlenecks you will work through next.