Signal4i · AI Economy · Enterprise · May 2026

The Internal FDE

The role the market can't fill already exists.

Last week, the Wall Street Journal confirmed what the data has been building toward for months. The AI economy is not eliminating technical roles — it is recomposing them. The role surviving the recomposition requires something the market cannot produce fast enough: deep process knowledge, agent fluency, and the ability to work embedded inside an organization's most critical systems. Aaron Levie, CEO of Box, named it the internal FDE. The question every regulated-industry organization is now asking is where to find one.

Read: The Tech Jobs That Are Safe From AI · WSJ →

The Data Has Been
Building Toward This.

The Wall Street Journal called it a tough time to be a software engineer. The data tells a different story. The sector is growing. The composition is shifting. The roles being eliminated and the roles being created are not the same roles. What the market is hunting for is something it cannot yet name at scale. In April, we named it in New Orleans. Read the PowerUp session →

+14.2%
IT & CS Job Postings
Year-over-year in April 2026. The sector is growing — not contracting. But the composition is shifting entirely.
43.1%
Senior-Level Share of Postings
Up from 38.8% a year ago. Entry-level fell from 8.1% to 7.4%. The floor dropped out of commodity roles. The ceiling on senior talent just got higher.
20%
Cloudflare Workforce Cut
1,100 jobs eliminated on a day of record revenue — $639.8M, up 34% YoY. CEO Matthew Prince called it an "agentic AI-first operating model." Internal AI usage surged 600% in three months.
The Pattern
Cloudflare. Block. Meta. Coinbase. Record revenue. Fewer humans. Not the same humans — different ones. The roles being shed are coordination, repetition, and support. The roles being hired require process knowledge, agent fluency, and the ability to work embedded inside an organization's most critical systems. That profile is what every regulated-industry organization is now competing to find.
The Paradox
The Same Companies Cutting Thousands
Are Simultaneously Posting More Jobs Than a Year Ago.
Amazon eliminated approximately 30,000 roles across late 2025 and early 2026 — then immediately announced plans to hire 11,000 software developers and interns in 2026. AWS CEO Matt Garman was direct: "We are hiring just as many software developers as we ever had inside of Amazon — demand for such talent is really accelerating." These are not contradictions. They are the same signal read from two directions. The workforce is being recomposed, not reduced. IT and CS job postings rose 14.2% year-over-year in April even as 113,863 tech workers were laid off in 2026 to date. Companies report a 92% increase in hiring for AI-related positions — with a 56% wage premium on those roles. The workers being laid off are largely not the workers being hired. The question isn't whether there are jobs. The question is which side of the recomposition you're on.

Amazon · PeopleMatters · May 2026 →    Big Tech Layoffs Analysis · Invezz →
92%
Increase in AI Role Hiring
Companies simultaneously cutting traditional roles and posting AI-adjacent positions at record rates.
56%
Wage Premium on AI Roles
High-demand AI engineering, MLOps, and agent architecture roles command a significant premium over displaced positions.
113K+
Tech Workers Cut in 2026
179 layoff events through May 13 — yet IT/CS postings rose 14.2% year-over-year in the same period.
Aaron Levie, CEO of Box, described the role his company
is now hiring for. The market doesn't know where to find them yet.
"Starting to hire and retrain for new agent engineering roles for internal functions to help get more powerful agents working well on critical business processes… It looks something like an internal FDE, whose job it is to wire up internal systems and get agents working with them effectively. The person will be extremely technical and capable of building secure, governed agents for internal workflows that connect to business systems… In some cases this person may understand the business process well enough to do it fully, but in most cases I expect them to work with the business directly in an embedded fashion… It's not about bringing automation to a job, but bringing automation to a process."
Aaron Levie — CEO, Box · LinkedIn · May 2026 →

FDE — Field Delivery Engineer — is a customer-facing role. Someone a vendor sends to a client to implement solutions on-site, embedded, hands-on. Levie is saying the new agent engineering role looks like that, but pointed inward at the company's own systems instead of outward at a customer. That inversion is the signal.

Companies are describing a role they need to hire using the vocabulary of a role they used to procure. They're trying to bring the implementation capability inside because the external market can't supply it fast enough — and because the process knowledge required is too proprietary to hand to a vendor. Every bank. Every manufacturer. Every healthcare organization. Levie said it plainly: they're all going to need this person. The problem is almost none of them know where to look.

What the Role Requires
Technical + Process, Embedded
Levie named it precisely: extremely technical, process-fluent, cross-functional, embedded in the business. Not a consultant who parachutes in. Someone who knows the system and the people who run it. ZipRecruiter confirms the shift — AI expertise postings now emphasize soft skills: communication and collaboration above all. The technical layer is table stakes. The human interface is the scarce ingredient.
The Sentence That Changes Everything
"Automation to a Process, Not a Job"
This is why most enterprise AI implementations fail. They automate the task as defined by the job description — which is a human artifact — instead of redesigning the underlying process the job was built around. The job is a wrapper. The process is the thing. The person who can unwrap decades of process logic and wire agents to the actual business beneath it is extraordinarily scarce. The organizations that find that person first will compound that advantage permanently.

This is the defining move of the Human Agentic state — the transition point where organizations stop augmenting humans and start designing processes around agents governed by humans. Most organizations are still trying to automate jobs. The internal FDE automates processes. That distinction is the entire gap.

Signal4i · The Coordination Tax · Silver Lake Series →

This Is What the Organization
Looks Like on the Other Side.

The agentic org is not a future concept. It is being built right now — in public, on earnings calls, in headcount decisions. The companies that have crossed the wall are already telling us what it looks like. The internal FDE is the architect of that transition.

The Agentic Org · Architecture in Production
One human. Intent set. Authority held. Agents executing beneath — governed, auditable, compounding.
Live Proof · Block / Jack Dorsey
40% Restructured Around One Principle
Block eliminated 40% of its workforce in February 2026 — not because the work disappeared, but because the ratio changed. Dorsey's stated model: one human, directly responsible, agents executing beneath them. Not a team. Not a department. One person with agent leverage. That is the agentic org architecture in production. The internal FDE is the person who designs and governs the agent layer that makes the ratio possible.
Live Proof · Cloudflare / Matthew Prince
Record Revenue. 1,100 Fewer People. More Employees Expected in 2027.
Cloudflare cut 20% of its workforce on the same day it reported $639.8M in revenue — up 34% year-over-year. CEO Matthew Prince called it an "agentic AI-first operating model" and expects more total employees by end of 2027 than at any point in 2026. Internal AI usage surged 600% in three months. This is not downsizing. It is recomposition at speed. The roles that survived: customer-facing and code-creating. The coordination layer between them: agents.

TechCrunch · Cloudflare Earnings →
The Architecture of the Agentic Org
Processes, Not Jobs
Job descriptions describe what humans do. The agentic org redesigns the underlying process — then decides which steps belong to agents and which require human judgment. The job is eliminated. The process survives and accelerates.
Knowledge Encoded, Not Carried
In the pre-agentic org, institutional knowledge lived in people. When they retired, it left with them. In the agentic org, knowledge is encoded as agent skills — governed, auditable, compounding. The internal FDE is the person who makes that encoding happen.
Human Authority at the Top
The agentic org does not remove humans. It elevates them. Humans set intent, govern outcomes, and handle the exceptions agents cannot resolve. The ratio shifts. The responsibility doesn't. This is the Human Agentic state — and it requires people capable of operating at that level.

Signal4i · 2,000 CEOs Confirmed It · Org Readiness Brief →    Human-Agentic Shift Visual →

None of These Voices Are From
the IBM i Community.
Every One of Them Is Describing It.

Levie doesn't know this platform exists. The Wall Street Journal didn't mention it. ZipRecruiter isn't tracking it separately. And yet — every requirement they named for the internal FDE maps precisely to the profile of the IBM i practitioner. The role the market is scrambling to create has been running inside IBM i organizations for thirty years. The question is whether the people who hold it recognize it before the window closes.

The Knowledge Asset
Why the Rule Has That One Exception
Why the pricing rule has that one exception. Why the month-end job runs in that exact sequence. What the error code actually means versus what the docs say. Which customer relationships require human judgment. Where the bodies are buried in thirty years of business logic. This is the process knowledge Levie says his company will struggle to hire. IBM i practitioners hold it without realizing it has a market value.
The Technical Asset
The Platform Is Already Agent-Ready
IBM i MCP Server. SQL Services. Mapepire. CL Commands exposed as callable tools. 500+ tools in 2026. The sovereign core — integrated database, object-based security, native job scheduler — was agentic-compatible before anyone called it that. The IBM i practitioner doesn't need to build the foundation. They're already standing on it.

And now the extraction layer exists. Project Bob — GA March 2026 — is an agentic AI development partner built specifically for IBM i modernization. It reads undocumented RPG, extracts the business logic, and encodes it into governed, auditable form. The process knowledge that lived in a person's head can now be surfaced, structured, and made operational. Project Bob doesn't replace the internal FDE. It is the tool that makes the internal FDE exponentially more productive.

Project Bob · bob.ibm.com →   IBM AI Coding Agent →   TechChannel · Bob GA →   Signal4i · KD Reduction Engine →
The Convergence
Domain Expert = System Author
Andrej Karpathy called it: "The bottleneck was never intelligence — it was the translation layer between knowing and building. That layer is collapsing." The IBM i practitioner has always been both. Knowing the business and building the system were never separate jobs here. That integration is exactly what the agentic org needs.
The Demand
Every Regulated Industry Is Hiring
Aaron Levie again, describing his customer base: "You can be sure that every bank, every pharmaceutical company, every healthcare company, every manufacturer is going to be hiring a ton of people to go implement agents." That is exactly the IBM i operating environment. The implementation demand hits harder here because the KD gap is wider and the internal talent to close it is scarcer.
The IBM i practitioner is not a legacy platform maintainer waiting to be replaced.
They are the internal FDE the market is desperately trying to hire — with thirty years of the business logic that makes the role worth anything at all.

The Knowledge Is the Asset.
Without AI Fluency, It's a Stranded Asset.

Victor Janulaitis, CEO of Janco Associates, said it plainly to the Wall Street Journal: engineers without AI-specific skills are still struggling to find work, even with many years of overall experience. This applies across every industry and every platform. Domain knowledge without the ability to operate at the agent layer does not cross the wall — it stops at it. The wall doesn't care how long you've been in the field. It only asks whether you've crossed it.

Read: The Knowledge Distance Problem → signal4i.ai
The Stranded Profile

Deep platform tenure. Thirty years of business logic. Irreplaceable institutional knowledge. Every one of the capabilities Levie is trying to hire.

But no AI fluency. No experience with agent frameworks. No vocabulary for describing what they know in terms agents can execute.

The knowledge walks out when they do. It was never encoded. It was never governed. And the window to change that is not permanent.

The Crossing Profile

Same platform tenure. Same business logic. Same institutional knowledge.

Plus: agent fluency. Able to surface knowledge, structure it as agent instructions, and architect the system that runs it.

The SME interview becomes the build. The knowledge stays and compounds. The bottleneck becomes the system author. This is the Human Agentic state — humans governing agents, not running every step. It is the profile the market is paying a senior premium for right now.

Signal4i · The KD Problem · Field Note →    Is There a Map? →

The difference between these two profiles is not talent. It is whether you crossed the wall.
The Three-State Transformation Model · Signal4i
STATE 01 Tech Upgrade AI tools deployed. Humans using AI as a better search. Process unchanged. Jobs automated at the margin. STATE 02 Human Augmented AI co-pilots active. Productivity gains visible but unscaled. Most organizations stall here. The KD Wall. THE KD WALL STATE 03 Human Agentic Agents govern process. Humans set intent, hold authority. Knowledge encoded. Internal FDE active. Agent-ability unlocked. 94% ADOPTION · 6% EBIT GAIN 72% SCALED · 33% GOVERNED THE DESTINATION
Framework: Signal4i Signal Stack v9.2 · Bottom stats: McKinsey GenAI Execution Research + HAI 2026  ·  Signal Stack v9.2 →

You Don't Have to
Figure This Out Alone.

The IBM i community has had an educational partner for decades. That partner is now building the path to the agentic layer — upskilling the veteran practitioner who holds the knowledge, and onboarding the next generation who carries the AI fluency. Together, they are the internal FDE team Levie is trying to hire as a single person. COMMON is where both halves meet.

01
The Crisis Is Already Named
COMMON's own research is unambiguous: more than 60% of IBM i professionals are aged 50 and above, and many are planning to retire in the next few years. The domain knowledge is not being replaced — it's walking out the door. Every retirement is a process encoded nowhere, governed by no one, running in someone's head for the last time. The urgency is not abstract. It is demographic.

COMMON · Bridging the IBM i Skills Gap · Feb 2026 →
02
The Upskilling Mandate
The veteran practitioner already has what Levie cannot hire: thirty years of process knowledge, business logic, and institutional memory. What remains is the agent fluency — the vocabulary, the frameworks, the practical experience of wiring what they know to what agents can execute. That is a learnable skill. COMMON is building the curriculum. The KD Wall is not a permanent barrier. It is a crossing. TechXchange Atlanta, October 2026, is the next on-ramp.
03
The Pairing Model
Levie is trying to hire one person who has everything — deep process knowledge and AI fluency in the same body. That person is extraordinarily rare. The IBM i community's answer is better. Pair the domain expert who holds thirty years of business logic with the AI-fluent next-generation practitioner who knows how to wire agents. Together they encode the knowledge, cross the wall, and build the internal FDE capability from the inside. COMMON is where both generations exist in the same room — and where the transfer becomes structural, not accidental.

COMMON · Veteran + Next-Gen Pairing Model →
04
The Pipeline
The next generation of IBM i practitioners is being built right now. The COMMON Education Foundation works directly with IBM's Power Systems Academic Initiative — connecting university students to the IBM i community through a skills-match program that pairs new graduates with employers. IBM committed to skilling 30 million people by 2030. COMMON is the IBM i community's access point to that investment. The pipeline exists. The question is whether your organization is positioned to receive what comes through it.

COMMON Education Foundation + IBM PSAI →
The COMMON Thesis
"The most valuable role in the AI economy requires two things the market cannot produce fast enough: thirty years of domain knowledge, and the fluency to wire it to agents. The IBM i community has the first. COMMON is building the path to the second. And the next generation coming through the pipeline brings both sides closer together."
Every bank, every manufacturer, every regulated-industry organization is about to compete for people who can do what Aaron Levie described — wire agents to critical business processes, embedded, process-fluent, technically capable. The IBM i community is the only place in enterprise computing where that capability can be built from existing assets rather than hired from a market that doesn't yet have them. The community is the competitive advantage. Upskilling activates the veteran. The pipeline brings in the next generation. COMMON is the infrastructure where both happen.

COMMON · common.org →    IBM PSA · IT Jungle →    IBM 30M by 2030 →

Imagine the person who spent thirty years watching good ideas die in the backlog. Defending systems nobody understood. Carrying knowledge nobody captured. Being told the platform was legacy, the skills were legacy, the future was somewhere else.

Last week,
the Wall Street Journal

described that person
as the most valuable new hire in the AI economy.
The process knowledge was always the asset.
The platform was always ready.
The window is open right now.
They had the slippers the whole time.
They just didn't know it yet.
The bifurcation is already visible in the data. Senior-level postings are surging. Entry-level is shrinking. Cloudflare cut 20% of its workforce on a record revenue day and called it an agentic AI restructuring. Block restructured 40% around a single principle: one human, directly responsible, agents executing beneath them. The midlevel manager layer is being replaced by the agent coordination layer. What survives is process knowledge plus agent fluency. That combination exists in concentrated form inside IBM i organizations — and nowhere in enterprise computing is the pairing of domain expertise and platform readiness more complete. The wall is real. The crossing is learnable. COMMON is the path — for the practitioner who is ready to act, and for the next generation stepping in to build alongside them.
Signal4i · Tracking what matters, for practitioners who are ready to act.