Skip to main content
Perspective:
Perspective
16 Jun 2026
23 minute read

The work chart vs. org chart: AI-driven organizational delayering and cross-functional role fusion

A perspective on operating model redesign, role fusion, and the entry-level talent shift

Vivek Kulkarni
Maya Bodan
Jamie Kilgour

Key takeaways

  • AI isn’t just accelerating tasks; it’s reducing the coordination and synthesis work that has historically required layers of management—and trained junior employees.
  • This allows organizations to fuse roles across functions and delayer traditional hierarchies, provided they maintain instrumented governance and controls.
  • Role fusion applies to every function, reframing organizational structure as a “work chart” vs. org chart, where tasks and processes—not people—are mapped to show how work gets done.
  • As managers shift to oversight/coaching and entry-level workers shift to monitoring/evaluation, organizations must redesign apprenticeship to avoid creating a capability gap in the coming years.

Every layer of management in your organization exists because, at some point, a human being was the most reliable way to gather information, interpret it, and move it to the next decision point—and AI is systematically eliminating that justification. By shortening the distance between workflow steps (sense → interpret → decide → execute → document → communicate), AI renders conventional handoffs and management layers unnecessary.

With this reduction in friction, managerial roles shift to orchestration and talent development1 while entry-level work moves beyond first-draft production to verification, evaluation, monitoring, and exception handling. “Delayering” the organization in this manner is not about cutting roles; it’s about faster decisions, less rework, stable quality, and measurable outcomes—all of which are validated by a scorecard of leading indicators.

Why layers exist (and why they’re under pressure)

Organizations don’t add layers because they enjoy bureaucracy. Hierarchies have grown out of deterministic workflows, where information scarcity and process control have justified escalation chains, reviews, steering committees, and “translation roles” across silos. In traditional operating models, these layers help manage friction, information latency, dependency load, uncertainty, and monitoring overhead through human relay.

Skip to description

Traditional hierarchy: Coordination by escalation

Hierarchical layers emerge when information scarcity and uncertainty are managed through sequential handoffs, which require human coordination and synthesis.

Why the hierarchy behaves this way
Information latency

Delayed, fragmented, or incomplete information forces people to escalate for clarity and decisions.

Dependency load

Work relies on other teams, systems, and approvals—creating queues and coordination overhead.

Quality uncertainty

Without visibility standardization, humans must check, double-check, and validate.

Monitoring overhead

Status, progress, and risk reporting consume time that could be spent on doing the work.

AI disrupts this pattern by introducing probabilistic workflows, where variability is expected and learning loops matter more than linear routing. In these systems, organizations anchor on outcomes and intelligent decision environments. Information is continuously captured from everyday artifacts (e.g., workflows, customer interactions, tests, incidents), enabling routine synthesis on demand. That minimizes manual routing and judgment at scale, enabling clearer decisions, fewer handoffs, and adaptive governance.

These capabilities render traditional layers optional, if not obsolete, giving leaders an opportunity to redesign roles and the ways in which work gets done.

Delayering teams: From hierarchy to AI-enabled network 

The structural implications of AI’s impact are not hypothetical: Deloitte’s 2026 State of AI in the Enterprise report found that one-third of organizations are already using AI to deeply transform core processes or business models, with another third actively redesigning key processes around AI.2

The organization of the future is not a boss-free one, however. In an AI-enabled operating model, the middle “relay” layers shrink—through redeployment and role redesign—as less human effort is required to gather, summarize, and route work across the system. Coordination (routing, summarization, first-pass verification, etc.) is increasingly automated, with humans adopting new, fused roles focusing on AI orchestration, coaching, oversight, and exception governance.

Skip to description

AI-enabled network: Expert-led and agile

Human engineers (HE) and supervisors (HS) manage a wider span of lightweight, real-time relay layers.

When AI absorbs routine coordination, what remains is the work that resists automation: resolving genuine ambiguity, governing exceptions that fall outside defined thresholds, and developing professionals who are learning in an environment where the old apprenticeship scaffolding no longer exists. A manager overseeing five people across a flattened workflow faces harder judgment calls, not easier ones.

This network model isn’t just for engineering. Role fusion can be applied to any artifact-heavy, coordination-driven work where AI reduces handoffs and enables more integrated roles with end-to-end ownership, such as:

  • Financial performance narratives;
  • Campaign production;
  • HR workflow; and
  • Service resolution.

The economics of the work chart vs. org chart: Delayering pays off in coordination costs 

Most AI discussions focus on task automation: drafting, summarizing, and generating content. Those gains matter, but the cost of coordination—the friction of routing work, interpreting signals, aligning decisions, and enforcing consistency—is where organizational delayering shines.

AI enables three fundamental shifts that help reduce coordination costs:

  1. Telemetry replaces manual reporting.
  2. Synthesis becomes continuous rather than meeting- or communication-driven.
  3. Verification becomes more scalable with explicit controls.

Instead of relying on calendar-based governance (like weekly status meetings), organizations can move to signal-based governance with continuous indicators and exception handling,3 lightening layers of management without losing control or quality.

The cost savings are significant: Research found that AI-assisted customer service agents were 15% more productive on average, with the biggest gains (34%) for the least experienced workers4—because AI was surfacing suggestions that senior agents previously provided informally in many cases. Similar patterns are emerging in professional services: the same State of AI report shows that organizations are redesigning core processes and business models around AI, suggesting coordination cost reduction extends beyond frontline service roles.5

Role fusion: What to expect and where to apply it

Role fusion is often seen in “forward-deployed” builder roles because the mechanics are more visible in software delivery: fewer handoffs, faster iteration, tighter accountability for outcomes.

In the sample work loop below, AI reduces coordination load in steps 3, 5, and 6, enabling one “forward-deployed” engineer to own more of the loop:

  1. Intake requests and needs
  2. Clarify context and restraints
  3. Execute, produce, and revise
  4. Validate quality and risk
  5. Document for traceability
  6. Communicate with stakeholders/reporting

But the title is not the point. “Forward-deployed” is best treated as an archetype: a broadly capable, outcome-oriented builder who can translate intent into a working solution with minimal handoffs. Similar roles in nontechnical domains include campaign producers, performance narrative owners, service resolution leads, and talent advisors. The label changes, but the principle stays the same: Broaden ownership and use AI to reduce handoffs and unite adjacent skills within a single role.

Role fusion applies to work that is repeatable, artifact-heavy, and coordination-driven. The early wins typically come from compressing “loop time”—the time it takes to move from question to decision to action—by removing unnecessary handoffs. Examples include:

  • Finance: A performance narrative owner can assemble evidence, draft reconciliations, generate variance commentary, and produce leadership-ready narratives while control functions focus on evidence retention, automated checks, and exceptions.
  • Marketing: A campaign producer can create audience-specific variants and synthesize performance insights while brand leadership focuses on guardrails.
  • HR and talent: Workflows similarly shift from administrative coordination to talent advising.
  • Customer operations: A resolution owner handles cases end to end—from triage through response—within defined escalation thresholds, reducing handoffs and queue time.

Role fusion and delayering are more feasible in lower-risk, reversible workflows with instrumented decision environments and explicit exception thresholds, where quality can be validated through observable signals and issues can be corrected quickly. Entry-level learning and development redesign is also a critical precursor. Without these conditions, coordination work reappears informally through shadow approvals and undocumented escalation paths.

They’re less feasible—and often constrained—where separation of duties is legally required or operationally critical (e.g., regulated authorizations, certain financial controls, security-sensitive approvals, safety-critical decisions, etc.). These environments typically use a “fused build, separated approve” approach, where AI accelerates preparation and first-pass work while approval boundaries remain explicit, auditable, and accountable by humans. Leaders should treat this as a design constraint, not an afterthought.

Organizations should also ensure AI-assisted workflows comply with data privacy, confidentiality, and intellectual property requirements, particularly when processing sensitive financial or personal data. The EU AI Act classifies workplace AI uses such as recruitment and performance evaluation as high-risk, requiring transparency, human oversight, and worker notification—reinforcing the “fused build, separated approve” principle as a regulatory baseline in this instance, not merely a design constraint.6

Role fusion in practice: The monthly close

The monthly financial close at a midsize technology company is a repeatable, artifact-heavy, coordination-intensive process—a prime opportunity for role fusion redesign, as illustrated below:

Before & After
Before & After
Dimension
Before
After

Distinct roles

3 (accountant, analyst, manager)

2 distinct role types (same headcount: narrative owner and review/governance)

Work loop

- Accountant reconciles accounts and flags variances. 

- Analyst investigates, pulls supporting data; drafts and edits narrative.

- Manager reviews and edits based on leadership context; assembles deck.

- Reconciliation via automated matching and flagging.

- Narrative owner  investigates flags; uses AI to create/edit draft and assemble deck.

- Manager reviews, handles exceptions, and provides coaching.

Handoffs per cycle

5, including loop repetitions when draft requires rounds of editing

1 (final review)

Days to leadership narrative

6–8 working days, roughly 40% spent on coordination rather than analysis.

3–4 working days

Manager time on synthesis

About ~2 days/cycle

Less than 0.5 days/cycle

Separation of duties

Informal (embedded in workflow)

Explicit (threshold-based sign-off, audit trail)

Illustrative example, based on a representative midsize technology-company close process; actual results will vary by control environment, system maturity, and data quality.

Key learnings

No roles are eliminated. Responsibilities are redistributed so that each person operates at a higher level of judgment:

  • Accountant and analyst responsibilities are combined into a single “performance narrative owner,” while reconciliation shifts upstream through automation—freeing up human capacity for exception resolution and source-data integrity. During close week, the narrative owner investigates flagged items and uses an AI assistant to generate and edit a draft based on variance data, prior-period narratives, and their own judgment.
  • The finance manager shifts from rewriting drafts to providing exception governance, reviewing the final package against control checkpoints, and providing developmental feedback.  

Guardrails matter as much as speed gain. Role fusion did not eliminate oversight in this example; editorial oversight turned into a set of explicit, auditable checkpoints. Evidence trails are retained automatically, variance thresholds trigger mandatory review, and AI-generated drafts are treated as inputs to human judgment—not substitutes for it.

AI doesn’t flatten organizations uniformly. It compresses coordination-heavy workflows in areas where handoffs are by-products of information latency. At the same time, it increases managerial complexity in high-ambiguity domains. As a result, the enterprise retains a hierarchical backbone of capabilities and accountability while selectively flattening certain execution stacks where AI reduces coordination friction.

Examples of role fusion beyond engineering
Examples of role fusion beyond engineering
Function
Traditional fragmentation
Fused outcome role enabled by AI
Non‑negotiable guardrails

Finance

Reconcile > analyze > narrative > reporting

Performance narrative owner / close outcome lead

Evidence trails; approvals; segregation of duties

Marketing

Research > messaging > variants > reporting

Campaign producer

Brand standards; claims substantiation;    approvals where required

HR and talent

Case intake > documentation > scheduling > policy guidance

Talent advisor

Fairness; privacy; accountable human decisions; compliance with regulations

Customer operations

Triage > knowledge  > escalation  > response

Resolution owner

Escalation thresholds; safety checks; customer commitment controls

PMO/delivery ops

Status collection  > deck prep  > meeting cadence

Outcome telemetry lead

Source-of-truth artifacts; exception governance;   auditability

Analytics

Data pull  > analysis  > narrative  > stakeholder comms

Insight-to-action owner

Data lineage; reproducibility; decision transparency

Apprenticeship redesign: How roles will change

AI is adept at the “starter tasks” that have historically trained junior professionals: drafting, summarizing, analyzing, and producing initial drafts of content like reports and presentations. But AI doesn’t eliminate early-career roles; it simply changes the work mix: As AI absorbs those tasks, junior work shifts toward validation, testing, monitoring, evaluation, and exception handling.

How, then, will juniors grow their capabilities? Particularly in flat org structures with fewer relay layers to catch errors and fewer people troubleshooting under ambiguity, challenging outputs, and owning outcomes end to end? Instead of “learning by doing,” juniors learn by evaluating. Arguably, they must become even more capable than they are today.

Skip to description

The entry-level shift: From production to judgment

Swipe to see the shift from the old model to the new.

The imperative to redesign apprenticeship is an opportunity to build a truly AI-enabled workforce from the ground up. Three tactics for strengthening capability development in flat org structures include:

1. Verification residencies: Junior professionals spend structured rotations across workflow types and teams, auditing AI-generated outputs against source material and quality standards.

Example: A marketing associate reviews AI-drafted customer emails and social posts against brand guidelines, factual accuracy, and tone—scoring each piece, flagging what’s off, and explaining their decisions and corrections in writing.

Why it works: Learning is embedded in the verification itself: Understanding why an output works or doesn’t work builds the same judgment that writing from scratch used to, perhaps faster. Difficulty increases over time, from straightforward fact-checking to ambiguous tone and strategy calls.

2. Dual-draft reconciliation. For high-learning-value tasks, a junior evaluates two independent drafts—one they wrote and one written by AI—identifying divergences and deciding which approach is stronger.

Example: A junior project manager drafts a stakeholder update and compares it to the AI version.

Why it works: The reconciliation process teaches them not only to write updates but also to evaluate effectiveness—a skill seldom cultivated in entry-level apprenticeships. This tactic is best used selectively for tasks where multiple valid approaches exist and should comprise no more than 20% to 30% of the workload.

3. Exception-handling rotations. Junior professionals rotate through roles where AI has flagged anomalies or low-confidence outputs—the cases that resist automation and require judgment under ambiguity.

Example: A junior at a help desk handles tickets the AI triage system couldn’t confidently route (e.g., a customer complaint that spans three products or a request that doesn’t fit a template).

Why it works: Each exception is a learning case. In HR, it might be a leave request that spans multiple policy exceptions or a compensation inquiry with no clear precedent. Difficulty should be calibrated and increased over time, with structured debriefs by a senior professional.
 

Delayering does not eliminate management work; it increases the complexity and stakes of the role while reducing routine activities (e.g., status collection, meeting minutes).7 The managerial center of gravity shifts from status synthesis and escalation routing toward:

  • Designing intelligent decision environments;
  • Governing uncertainty thresholds;
  • Coaching AI-augmented teams; and
  • Ensuring teams stay focused on high-level thinking tasks instead of defaulting to low-level executional tasks.

If organizations delayer teams without upgrading these capabilities, they will likely either re-create bureaucracy informally (shadow approvals) or absorb higher operational risk.

As organizations delayer and flatten, apprenticeship redesign becomes more than a talent initiative—it’s a structural risk-control mechanism. Without it, a capability gap could form in just a few years, along with other enterprise risks, such as:

  • Gaps in judgment development;
  • Shadow review layers and rework;
  • Higher operational risk and lower quality; and
  • A weaker pipeline of future managers and leaders.

Apprenticeship tactics at a glance

All three tactics require the same organizational commitment:

  • Dedicated capacity: Development time must be explicitly budgeted rather than absorbed by delivery tasks.
  • Outcome-based metrics: Progress should be measured by observable indicators instead of tenure-based proxies.
  • Leadership accountability: Senior professionals must be held responsible for coaching.

Without these commitments, apprenticeship redesign dissolves on contact with delivery pressure—and the organizations most at risk are the ones that have already delayered.

Tactic
What the junior does
What it builds
Best suited for

Verification residency

Audits AI outputs against source data; scores and documents errors

Domain knowledge, pattern recognition, quality judgment

Repeatable,   artifact-heavy work (finance, audit, reporting)

Dual-draft    re-conciliation

Produces independent draft, then reconciles against AI version

Analytical skill, evaluative judgment, comparative reasoning

Strategy, analysis,   design tasks with multiple valid approaches

Exception-handling   rotation

Resolves cases AI flagged but couldn’t confidently handle

Ambiguity tolerance, contextual judgment, de-escalation

Operations, customer   service, compliance, audit

The blueprint: Delayer safely, fuse roles selectively 

Organizations that succeed in flattening hierarchies tend to follow a similar sequence:

  1. Map the work. Before touching the org chart, identify candidate workflows. Outline handoff points and bottlenecks, and make decision rights explicit.
  2. Instrument governance. Identify telemetry sources, network traces, and quality signals for each workflow—these will replace manual status collection and approval chains.
  3. Fuse roles selectively. Where the risk profile supports it, combine role responsibilities to expand end-to-end ownership across the work loop.
  4. Add guardrails. Add controls, approvals, and separation where necessary.
  5. Rebuild the career ladder. Redesign entry-level development with intention, weaving in coaching and rotation-based learning.

A common failure is delayering teams while leaving workflows unchanged. If you automate a broken process, you scale the breakdown: The coordination work returns—just in a less visible, less auditable form. Sequence matters: governance before role fusion, role fusion before guardrails, etc.—each step removes a safety net that the next step depends on.

Organizational delayering measurement

A delayering initiative should be evaluated on outcomes and risk, not org-chart aesthetics. Successful delayering occurs only when speed improves without quality degradation: Organizations that delayer without building managerial capabilities will tend to re-create the bureaucracy they eliminated—just informally, through undocumented approvals and ad hoc escalation paths. Signals of productivity beyond a flatter org chart include:
Decision latency

Less time from issue raised to decision | Escalations are fast and evidence-based | Less waiting for approvals

Cycle time

Less end-to-end time from idea to outcome |Less queue time between steps | More finished work per quarter

Rework and exceptions

No spike in defects/reversals | Exceptions are visible and handled quickly | Controls are embedded, not bolted on

Quality risk signals

Stable or improving incidents/audit findings | Traceable approvals where required | Monitoring catches drift early

Span-of-control health

Managers spend more time coaching | Manager load is sustainable | Attrition and burnout do not rise

Value realized

Outcomes measured vs. targets | Benefits track to accountable owners | Less activity theater, more results

A turning point for organizational structure

Organizational delayering is not a slogan. It’s a form of operating model redesign that enables teams to operate with fewer coordination-heavy relay layers—but only when organizations replace hierarchies with instrumented governance, explicit decision rights, auditable controls, and deliberate talent development.

Within three years, delayering and role fusion are likely to separate market leaders from laggards just as digital transformation did more than a decade ago—and with the same early skepticism followed by rushed adoption. Organizations that wait may be forced to apply it under competitive pressure, which is when most redesigns fail. Those that use it as a mere headcount lever are likely to re-create bureaucracy in new, less visible forms. The ones that move intelligently now are building muscle memory for an AI-enabled operating model with fewer coordination-heavy relay layers, greater judgment, and sustainable speed.

Endnotes

1. Sue Cantrell et al., “Is there still value in the role of managers?,” Deloitte Insights, March 24, 2025.
2. Jim Rowan et al., State of AI in the Enterprise 2026: The untapped edge, Deloitte, January 2026.
3. Jim Rowan et al., “The agentic reality check: Preparing for a silicon-based workforce,” Tech Trends 2026, Deloitte Insights, December 10, 2025.
4. Erik Brynjolfsson, Danielle Li, and Lindsey Raymond, “Generative AI at work,” Quarterly Journal of Economics 140, no. 2 (May 2025): pp. 889–942.
5. Rowan et al., State of AI in the Enterprise 2026: The untapped edge.
6. European Parliament, Regulation (EU) 2024/1689 (EU AI Act), June 13, 2024.
7. Cantrell et al., “Is there still value in the role of managers?”

Did you find this useful?

Yes
No

Thanks for your feedback

Your feedback is important to us

To tell us what you think, please update your settings to accept analytics and performance cookies.

Need help updating cookies?

Get in touch

Vivek Kulkarni

Vivek Kulkarni

US AI Transformation Leader
Maya Bodan

Maya Bodan

Managing Director | Human Capital Services
+1 415 783 4000
Jamie Kilgour

Jamie Kilgour

Managing Director
12078005318

Related insights

Deloitte AI Institute - AI Insights

The Deloitte AI Institute helps organizations transform through cutting-edge AI insights and innovation by bringing together the brightest minds in AI services.
Collection