For decades, managers have responded to growth with a familiar sentence:

“We need another person.”

Sales volume increases.

Hire another salesperson.

More customers.

Hire another support agent.

More invoices.

Hire another accountant.

More employees.

Hire another HR coordinator.

More reports.

Hire another analyst.

Workload goes up.

Headcount follows.

It became so normal that most managers stopped thinking of hiring as a design decision.

There is work.

Therefore we need a worker.

AI is beginning to break that assumption.

Because the next unit of productive capacity inside a company may be:

a person,

software,

automation,

an AI assistant,

an AI agent,

an external provider,

or some combination of all six.

That changes hiring from a simple workforce question into something much more interesting:

What is the best way to buy the capacity we need?

Not:

“Can AI replace this employee?”

That’s too crude.

The better question is:

“If this company needs more capacity, what combination of people and technology should we add?”

That is a very different conversation.

And I think CEOs, CFOs, CHROs and managers will be having it increasingly often.

Before You Approve the Next Headcount Request

Imagine a department head says:

“We’re overloaded. I need three more people.”

Historically, leadership might ask:

How much will they cost?

Can we afford it?

How quickly can HR hire them?

AI adds several questions.

What work is creating the overload?

Why does that work exist?

Which tasks require a person?

Which tasks could disappear?

Which could be automated?

Could AI increase the existing team’s capacity?

Could an agent own part of the workflow?

Would one new employee + automation outperform three new employees?

Could the company grow without adding the position at all?

These questions do not automatically lead to:

“Don’t hire.”

Sometimes the correct answer is still:

“Hire immediately.”

But AI means the decision deserves more analysis than before.

The Question Is Not Human Versus Machine

This is where most AI workforce discussions go wrong.

They become ideological.

One side says:

“AI will replace everyone.”

The other says:

“AI will never replace humans.”

Neither is particularly useful to a CEO deciding next year’s operating budget.

Businesses don’t operate through philosophical absolutes.

They allocate resources.

A warehouse uses:

people,

forklifts,

software,

conveyors,

robots.

Nobody asks:

“Will forklifts replace warehouse employees?”

The useful question is:

“Which resource should perform which work?”

The knowledge economy is entering the same transition.

The difference is that until recently, many cognitive tasks had only one realistic execution engine:

a person.

AI changes that.

Research.

Classification.

Monitoring.

Drafting.

Coordination.

Analysis.

Customer interaction.

System updates.

Routine decisions.

Some of this work can increasingly be executed without adding another human hour.

So workforce planning needs another category.

Your Workforce Is Becoming More Than Your Employees

Traditionally, companies think about workforce capacity through:

full-time employees,

part-time employees,

contractors,

consultants,

outsourcing.

Now add:

digital labor.

Not as marketing language.

As actual productive capacity.

Imagine a company has:

200 employees.

But also:

a sales qualification agent,

a finance reconciliation agent,

an employee-support agent,

a customer-service agent,

a reporting agent,

several development agents,

marketing automations,

and dozens of AI assistants used by employees.

What is the size of that workforce?

200?

Perhaps legally.

Operationally?

The answer becomes much less obvious.

McKinsey describes this emerging structure as a hybrid workforce, where people and intelligent systems work side by side and organizations must determine where to deploy both human talent and agents to create the most value. [1]

That means workforce planning itself needs redesign.

Headcount Is Becoming an Incomplete Metric

CEOs love headcount.

It’s simple.

How many employees?

How many did we hire?

How many are leaving?

What is revenue per employee?

What is payroll?

But imagine Company A employs:

1,000 people.

Company B employs:

700 people + 5,000 agents performing portions of internal workflows.

Are these economically comparable companies?

Not really.

You may eventually need to measure something closer to:

productive capacity

rather than only:

number of humans employed.

I’m not suggesting a universal formula yet.

The market is too early.

But the direction matters.

If agents can perform increasing amounts of commercially useful work, workforce economics can no longer be understood solely through headcount.

Do Not Translate “Agent” Into “Employee”

There is another mistake.

Some companies will try to calculate:

“One AI agent equals 3.7 employees.”

That sounds wonderfully precise.

It is usually nonsense.

People and agents are not interchangeable units.

A human can:

understand unexpected context,

build relationships,

learn across messy situations,

take responsibility,

negotiate,

work across unclear objectives,

recognize organizational politics,

adapt when the environment changes completely.

An agent may be able to:

perform certain workflows,

monitor continuously,

process enormous volumes,

coordinate systems,

repeat tasks consistently,

operate 24/7.

Different strengths.

The question isn’t:

How many employees does this agent replace?

It is:

What capacity does this operating model create?

That’s more useful.

Sometimes You Should Hire the Employee

Let’s make this clear before continuing.

There are plenty of situations where hiring a human is the correct business decision.

Especially when the work depends heavily on:

relationships,

trust,

ambiguous judgment,

leadership,

negotiation,

physical presence,

accountability,

creativity,

complex problem-solving,

institutional knowledge.

Suppose your company is expanding into Saudi Arabia.

You need someone to build relationships with strategic accounts.

AI can:

research prospects,

prepare account intelligence,

translate information,

manage follow-up,

support proposals.

Should you deploy an autonomous AI Country Manager?

Probably not.

You need a person.

Give that person extraordinary AI capability.

But hire the person.

The human relationship is part of the work.

Sometimes You Should Not Hire Anyone

Now imagine a company receives 5,000 supplier invoices per month.

Volume increased.

Finance requests two additional employees to:

extract invoice information,

check purchase-order numbers,

match suppliers,

identify discrepancies,

update records,

route exceptions.

This may be legitimate work.

But before approving two salaries forever, ask:

Why does a human need to touch every invoice?

Perhaps the better operating model is:

software handles structured matching,

AI extracts unstructured data,

rules validate known conditions,

an agent coordinates the workflow,

humans investigate exceptions.

Now increased volume may not require increased headcount at all.

This is not replacing two people.

The positions never needed to exist.

That’s a very important distinction.

The Best Time to Redesign a Job Is Before You Hire for It

Companies often introduce AI into existing jobs.

That’s difficult.

The employee already exists.

Responsibilities exist.

Team structures exist.

Expectations exist.

Processes exist.

Political considerations exist.

Now imagine a manager requests a new role.

That is an opportunity.

Nothing has been institutionalized yet.

Before creating the job description, ask:

If we designed this role from zero today, what would it contain?

Maybe the department says it needs a:

Marketing Coordinator.

Why?

To:

prepare reports,

update campaign data,

coordinate requests,

schedule content,

research competitors,

draft basic copy,

send reminders.

Almost the entire role consists of coordination and information work.

Perhaps the company does not need a Marketing Coordinator as traditionally designed.

It may need:

better automation

an AI-enabled workflow

a more senior marketer with greater capacity.

Or perhaps after redesigning the process, the company still needs the coordinator—but the job changes.

That’s fine too.

The purpose of the exercise is not:

eliminate the job.

It is:

stop hiring for work that technology has already changed.

The Hire → Automate → Agent Decision

When additional capacity is requested, there should be at least four possible answers.

1. Eliminate the Work

Does it need to exist?

This remains the first question.

As we argued in Article 04:

Don’t automate what you haven’t challenged.

If the report nobody reads is creating workload, don’t hire someone to produce it faster.

Delete it.

2. Automate the Work

Is it:

predictable,

repetitive,

rule-based,

structured?

Use conventional automation when possible.

You don’t need an agent for:

sending reminders,

updating simple fields,

moving known data,

triggering predictable workflows.

Use the simplest technology that reliably solves the problem.

3. Deploy an Agent

Does the work involve:

interpretation,

multiple steps,

unstructured information,

monitoring,

system interaction,

coordination,

bounded decisions?

Now agentic AI becomes interesting.

4. Hire a Person

Does the work primarily create value through:

judgment,

relationships,

leadership,

accountability,

deep expertise,

ambiguity?

Hire.

Then use AI to increase that person’s leverage.

This should become a normal management decision tree.

Brightery’s View of Automation Makes the Distinction Useful

Brightery’s work on automating company operations starts from an important operational principle:

automation is not simply replacing people.

It is removing repetitive work, delays, duplication and friction from the company’s operating system.

That’s the right starting point.

When the workflow becomes more efficient, headcount decisions become clearer.

You can see which work genuinely deserves human capacity.

And which work existed simply because systems were disconnected.

Ask What the Employee Would Actually Spend Their Day Doing

This is an extremely useful test.

Manager:

“I need another person.”

Fine.

Write the proposed employee’s week.

Not the polished job description.

The actual 40 hours.

For example:

10 hours — prepare reports.

8 hours — update CRM.

6 hours — customer follow-up.

5 hours — research.

4 hours — scheduling/coordination.

3 hours — meetings.

2 hours — analyzing problems.

2 hours — customer relationships.

Now look at it.

You are about to hire an entire human being because of a role where:

33 of 40 hours

may contain tasks that could be:

eliminated,

automated,

or augmented.

Maybe the right answer is still hire.

But perhaps the role should be redesigned around the seven hours where human value is highest.

That might allow one person to do what would previously require two.

That’s workforce leverage.

The Human Should Not Be the Automation

There is a type of employee found inside almost every growing business.

Their real job is:

open System A.

Find information.

Copy it.

Open System B.

Paste.

Check email.

Send reminder.

Update spreadsheet.

Ask manager.

Wait.

Follow up.

Create report.

Do it again.

We mentioned this earlier in the series:

The company is using a person as an API.

It happens because humans are flexible.

When systems don’t connect, companies hire someone to connect them manually.

When information is fragmented, companies hire someone to find it.

When processes are unclear, companies hire coordinators.

When reports are manual, companies hire analysts.

As the company grows, this produces entire layers of human middleware.

AI and automation provide an opportunity to remove some of that structural inefficiency.

Not because the people lack value.

Because the organization is wasting their value.

Growing Companies Should Be Especially Careful

At 20 employees, many processes work through conversation.

Need something?

Ask Ahmed.

Customer issue?

Sara knows.

Invoice problem?

Call finance.

At 100 employees, those informal mechanisms become harder.

The natural response is:

more coordination roles.

More administrators.

More middle management.

More systems.

More process.

More people moving information between people.

This is one reason companies often become slower as they grow.

AI creates a different possibility:

Scale coordination without scaling coordination headcount at the same rate.

That’s potentially enormous.

A company might grow:

customers 100%

revenue 100%

transactions 100%

while administration grows only:

30%.

That changes economics substantially.

This is what AI workforce planning should aim for.

Not simplistic layoffs.

Operating leverage.

AI May Reduce Hiring Before It Reduces Employment

This distinction deserves much more attention.

Imagine a support team has:

50 employees.

Business volume is expected to grow 40%.

Traditionally:

the company may need 20 more people.

Now AI agents handle routine requests and help employees resolve complex requests faster.

The business grows 40%.

Support headcount grows from:

50 → 56.

Did AI eliminate 14 jobs?

Not exactly.

Those people were never hired.

But economically:

the company avoided substantial future cost.

This is a much more likely near-term pattern in many businesses than dramatic replacement of existing teams.

And it’s one reason AI-related headcount impact is more complicated than headlines suggest.

Gartner cautioned in January 2026 against assuming that AI productivity translates directly into straightforward workforce reductions; it expects AI to shift the composition of headcount and create new roles even as some forms of work decline. [2]

That is a more realistic framework.

Cost Avoidance Is Real ROI

Finance teams should distinguish:

Cost Reduction

Existing spending disappears.

From:

Cost Avoidance

Future spending never appears.

Both matter.

Imagine an employee costs:

AED 240,000 annually fully loaded.

The business expected to hire:

Future payroll increase:

AED 2.4 million.

AI-enabled capacity means it hires:

four.

Avoided annual future cost:

AED 1.44 million.

That is real economic value.

But don’t call it:

AED 1.44 million savings

if the money was never in the current cost base.

Call it:

AED 1.44 million cost avoidance.

CFOs will appreciate the distinction.

The Agent Is Not Free Labor

Now we need to avoid the opposite mistake.

Some executives imagine:

Employee:

AED 300,000 annually.

Agent:

AED 500/month.

Obviously choose the agent.

Not so fast.

The agent may require:

software,

models,

tokens,

orchestration,

integration,

data infrastructure,

development,

monitoring,

security,

governance,

human review,

maintenance,

support,

exception handling.

There is also risk.

One poorly designed agent can create errors at machine speed.

So the real comparison is not:

Salary vs AI subscription.

It is:

Total cost of human capacity vs total cost of agentic capacity.

That is a much more serious analysis.

Calculate Total Human Cost Properly

Salary is not the full employee cost.

Depending on the company and market, there may also be:

recruitment,

visa,

benefits,

insurance,

office,

equipment,

software,

training,

management,

HR,

turnover,

onboarding,

paid leave,

administrative support.

Likewise, there are benefits not captured by salary:

institutional knowledge,

relationships,

adaptability,

leadership potential,

team contribution.

Don’t reduce the employee to payroll.

Calculate Total Agent Cost Properly

Likewise, include:

platform licenses,

model consumption,

infrastructure,

integration,

development,

monitoring,

security,

evaluation,

data preparation,

human supervision,

governance,

maintenance,

vendor risk,

failures.

The economics may still be excellent.

But the comparison needs to be honest.

Gartner’s June 2026 research warns that AI is not simply reducing workforce costs; it is moving costs into new categories that many businesses do not yet track properly. [3]

That observation should be written in large letters inside every AI business case.

The Hire-or-Agent Framework

When deciding how to add capacity, I would score six dimensions.

1. Volume

How much work exists?

Low volume may not justify automation.

Extremely high volume makes machine capacity more attractive.

2. Variability

How predictable is the work?

Same process every time?

Automation/agents become easier.

Completely different situation every time?

Human advantage increases.

3. Judgment

Does the work require genuine contextual judgment?

Not merely:

reading information.

But understanding:

trade-offs,

people,

uncertainty,

unspoken context.

Higher judgment generally pushes toward human ownership.

4. Relationship Value

Is the relationship itself part of the product?

Think:

enterprise sales,

leadership,

advisory,

luxury service,

healthcare,

negotiation.

If yes, human capacity remains important.

5. Risk

If something goes wrong, what happens?

Low-risk routine activities can support higher autonomy.

High-risk irreversible decisions require stronger human control.

6. Scalability

How quickly might workload increase?

Agentic capacity can often scale differently from human teams.

This becomes particularly attractive when volume is highly variable.

Then Add Two Practical Constraints

7. Data Readiness

Can AI actually access reliable information?

If not, your beautiful agent architecture may fail immediately.

8. Integration Readiness

Can the agent act inside the systems where the workflow happens?

If not, a human may remain trapped in the middle.

This is why workforce strategy is increasingly connected to technology architecture.

A Simple Decision Matrix

That final row may create the most savings.

Example: Customer Service

The business is growing.

Support volume increased 50%.

Manager asks for:

10 additional agents.

Traditional decision:

hire.

Let’s redesign.

Suppose:

60% of requests are routine:

order status,

appointment confirmation,

password issues,

standard policy questions,

simple changes.

20% require moderate interpretation.

20% are complex.

New model:

Routine 60% → AI agent.

Moderate 20% → AI handles most work, employee reviews or resolves exceptions.

Complex 20% → human support specialists.

Now perhaps the company hires:

three people,

not ten.

But those three may need to be:

more capable,

better paid,

better trained.

Workforce cost doesn’t disappear.

It changes shape.

That’s a pattern I expect to see repeatedly.

AI Could Make Some Human Roles More Valuable

This is the part cost-cutting discussions miss.

If agents handle:

routine support,

then human support employees increasingly handle:

angry customers,

complex situations,

retention risks,

exceptions,

high-value relationships.

The job becomes harder.

The person may need:

better judgment,

more authority,

stronger communication skills.

So while total headcount might grow more slowly, the value of remaining human roles may rise.

That means some companies may need to spend:

more per employee

while spending:

less per transaction.

That’s not contradictory.

That’s operating leverage.

Example: Sales

Sales director asks for:

five additional SDRs.

Why?

To:

research accounts,

identify contacts,

qualify leads,

send outreach,

follow up,

update CRM,

schedule meetings.

Now apply the framework.

Research?

AI.

Enrichment?

Automation.

Qualification?

AI + rules.

Routine follow-up?

Agent within boundaries.

CRM updates?

Agent.

Scheduling?

Automation.

Discovery?

Human.

Relationship?

Human.

Negotiation?

Human.

What does the company actually need?

Maybe not five traditional SDRs.

Maybe:

one or two highly capable sellers

agentic sales infrastructure.

Or maybe lead volume genuinely requires five people.

The point is:

you can no longer know until you redesign the work.

Udjat’s work on AI Marketing Automation UAE and How AI Can Change Marketing for Business in Dubai already points in this direction: AI can support qualification, personalization, analysis and automated follow-up, while people remain central to strategy and meaningful customer interaction.

That’s exactly the workforce redesign happening here.

Example: Finance

CFO says:

“Transaction volume is growing. We need three accountants.”

What do they need to do?

If the answer is:

reconcile,

match,

extract,

check,

prepare,

route,

follow up,

then agentic/automation capacity deserves investigation.

If the answer is:

financial control,

business partnering,

tax strategy,

planning,

capital allocation,

complex exception resolution,

the human case becomes stronger.

This could lead to a finance department with:

fewer people processing

and

more people interpreting.

Different workforce.

Better use of talent.

Example: HR

The company grows from:

200 → 500 employees.

Historically, HR administration may increase proportionally.

More:

questions,

documents,

leave requests,

onboarding,

coordination,

interviews,

scheduling.

But does HR headcount need to grow 150%?

Maybe not.

Brightery’s AI HR Agent demonstrates how employee requests, onboarding, common questions and workflow coordination can be automated or AI-assisted.

That can let HR add capacity where it matters:

talent strategy,

leadership development,

employee relations,

culture,

workforce planning.

Instead of adding another person primarily to answer:

“How many leave days do I have?”

You make the answer available automatically.

Example: Marketing

Marketing teams are already feeling something interesting.

AI dramatically increases production capacity.

One person can generate:

more copy,

more designs,

more variants,

more research,

more analysis.

So does marketing need fewer people?

Not necessarily.

Because content abundance creates different needs.

More importance may shift toward:

positioning,

creative judgment,

channel strategy,

customer research,

editorial direction,

performance analysis.

The production layer becomes cheaper.

The judgment layer becomes more valuable.

This is a good example of how AI does not simply shrink a department.

It changes where the department spends its human budget.

Hiring Should Move From Tasks to Capabilities

Traditional job requisition:

“We need someone to prepare reports and manage data.”

Future requisition:

“We need stronger commercial analysis capability.”

Then ask:

What creates that capability?

An employee?

AI?

Agent?

Software?

Combination?

The distinction sounds small.

It isn’t.

One is buying a person because there are tasks.

The other is designing a system to create an outcome.

That’s a much more mature workforce strategy.

Think in Outcomes per Dirham

Suppose two options.

Option A — Employee

Fully loaded annual cost:

AED 300,000.

Capacity:

2,000 productive hours.

Can handle:

complexity,

exceptions,

relationships.

Scales:

linearly.

Option B — Agentic Workflow

First-year cost:

AED 500,000.

Following annual operating cost:

AED 150,000.

Can process:

100,000 transactions.

Requires:

0.5 FTE supervision.

Which is cheaper?

The question is impossible to answer until we know:

what work?

what quality?

what risk?

what volume?

If the workflow runs:

1,000 times annually,

the employee may be cheaper.

At:

100,000,

the agent might dominate.

Economics depend on volume and task architecture, not ideological preference.

AI Changes Fixed and Variable Cost

This is an important CFO consideration.

Human capacity often adds cost in chunks.

Need more capacity?

Hire another person.

And another.

And another.

Agentic capacity may scale differently.

Some costs are fixed:

development,

integration,

governance.

Then usage adds variable cost:

model calls,

compute,

transactions.

This changes the shape of the cost curve.

A growing company should understand:

At what volume does agentic capacity become economically attractive?

There will be a break-even point.

Find it.

The Break-Even Question

Imagine a workflow costs employees:

AED 40 per case.

Agentic workflow total operating cost:

AED 8 per case.

But implementation costs:

AED 400,000.

Difference:

AED 32 per case.

Break-even volume:

400,000 ÷ 32

=

12,500 cases.

If you process:

2,000 cases per year,

the investment may not make sense.

If:

100,000,

very different.

This is how AI workforce discussions should sound.

Less:

“AI is the future.”

More:

“At what volume does the operating model make financial sense?”

Do Not Forget Management Cost

Hiring people creates management work.

Recruitment.

Onboarding.

Performance management.

Communication.

Coordination.

Meetings.

Career development.

That’s not bad.

People require leadership.

Agents require management too—but differently.

Agents need:

monitoring,

evaluation,

permissions,

governance,

updates,

exception review.

So the company does not eliminate management.

It changes what management manages.

Later in this series we’ll explore:

Who Manages 100 AI Agents?

Because this question will become much more important than it sounds today.

AI Agents Can Scale Quickly. Their Mistakes Can Too.

Suppose an employee makes a bad decision.

They can probably make:

several bad decisions per day.

An agent can potentially make:

thousands.

This is the inverse of agent scalability.

The same architecture that gives you:

cheap marginal execution

also gives you:

cheap marginal errors.

Therefore agent economics must include:

error rate,

impact per error,

detection speed,

reversibility,

control cost.

Cheap execution is not cheap when mistakes are expensive.

A Human Employee Has Something an Agent Doesn’t: General Adaptability

Suppose a business changes priorities suddenly.

An experienced employee can often adjust.

New customer issue?

Handle it.

Manager changes process?

Adapt.

Something completely unexpected?

Use judgment.

Agents are becoming more adaptable.

But many production agents remain best when:

the objective is bounded,

the available tools are known,

the environment is reasonably structured.

This is one reason humans remain extremely valuable in:

fast-changing,

ambiguous,

politically sensitive,

relationship-heavy environments.

Do not compare only cost per task.

Compare flexibility.

Employees Also Become Institutional Memory

A good employee knows things your databases do not.

Why the company changed a rule.

Which client prefers phone calls.

Why the obvious supplier is actually unreliable.

Which manager needs to be involved.

What happened during the project nobody documented properly.

Agents can increasingly capture and retrieve institutional knowledge.

But today, companies still possess enormous amounts of unwritten context.

Removing human capacity without first capturing that knowledge can create fragile organizations.

This becomes particularly dangerous when reducing experienced employees while relying on AI trained only on formal company documentation.

The AI Workforce Needs Knowledge Architecture

If you want agents to take meaningful work, ask:

What do they need to know?

Where does that knowledge live?

Can they access it?

Can they trust it?

PwC’s 2026 Middle East CEO research found strong regional confidence in AI adoption, yet only 22% of Middle East CEOs said their most-used AI tools had access to all relevant organizational documents and data; the GCC figure was only 16%. [4]

That matters enormously to workforce planning.

You cannot simply replace human coordination with an agent if the human currently holds half the process in their head.

Data and knowledge readiness determine how much agentic capacity you can actually deploy.

The UAE Is Not Simply Moving Toward Fewer Jobs

This is particularly important for the UAE.

The AI narrative often becomes:

AI → automation → fewer people.

The UAE labor-market data tells a more complicated story.

PwC’s 2026 AI Jobs Barometer found that the share of UAE job postings requiring AI skills increased from 1.0% in 2021 to 3.2% in 2025, with demand rising across every sector examined. [5]

The most AI-exposed roles also showed far greater skills change.

In other words:

AI isn’t only reducing demand for certain tasks.

It is increasing demand for different capability.

PwC’s 2026 UAE CEO research therefore recommends that CEOs redesign work around human-AI collaboration rather than cost reduction alone. [6]

That’s exactly the right framing.

AI Should Change Who You Hire

Suppose your company still needs another employee.

AI changes the profile you should look for.

Traditional employee:

good at producing.

Future employee may need to be better at:

defining,

judging,

directing,

integrating,

questioning,

improving,

managing agents,

handling exceptions.

Consider two analysts.

Analyst A can manually build an excellent report.

Analyst B can:

use AI to build it faster,

challenge the AI,

find the important insight,

understand the business,

recommend the right action.

Who becomes more valuable?

Probably Analyst B.

The skill premium shifts.

Hire People Who Can Multiply Agent Capacity

There is another emerging category:

people whose value comes partly from how effectively they can deploy AI capacity around them.

Imagine:

one operations manager

20 agents

versus

another operations manager

the same 20 agents.

One designs excellent workflows.

Monitors exceptions.

Improves agent behavior.

Allocates authority intelligently.

The other treats agents as chatbots.

Same technology.

Very different output.

That means the return on certain human hires may actually increase in the AI era.

A strong person can increasingly leverage a much larger system.

This Is Why the Best Employees May Become More Expensive

If one highly capable employee can manage a dramatically larger amount of work using AI, their economic leverage rises.

We may therefore see an interesting pattern:

fewer low-value coordination roles,

stronger demand for highly capable people,

higher compensation for scarce judgment,

and greater use of agentic capacity around those humans.

The future workforce may not simply be smaller.

It may become:

more polarized by leverage.

That has significant implications for hiring and development.

Don’t Remove the Junior Layer Without Thinking

There is one uncomfortable problem.

Many junior roles are built around:

research,

drafting,

data preparation,

basic analysis,

documentation.

AI is particularly strong at these tasks.

So a manager may reasonably conclude:

“Why hire juniors?”

Because eventually you will still need seniors.

Where do they come from?

If an organization removes the work through which people historically developed expertise, it needs a new development model.

This could include:

simulation,

apprenticeship,

structured AI-supervised learning,

deliberate manual practice,

earlier exposure to judgment-heavy work.

Deloitte’s research on the future agentic workforce warns that early-career roles may face especially significant disruption because many foundational tasks are highly exposed to automation. [7]

That’s not simply an HR problem.

It’s a future capability problem.

Hire vs. Agent Should Include the Five-Year Question

The cheapest choice this year is not necessarily the smartest choice.

Suppose eliminating junior hiring saves money for three years.

Year five:

you have no experienced pipeline.

That’s expensive.

Likewise:

perhaps hiring ten people for work likely to become mostly automated next year creates another problem.

Workforce planning needs to consider:

current capacity

and:

future capability.

Those are not always aligned.

The Agent Should Not Automatically Replace the Cheapest Employee

Another common assumption.

Where do companies automate first?

Often low-paid roles.

But economically, the greatest AI opportunity may exist around expensive professionals.

Imagine:

lawyers,

engineers,

consultants,

senior salespeople,

financial analysts.

If AI removes five hours per week from a highly paid professional and those hours become revenue-producing capacity, the economics can be enormous.

So don’t rank automation opportunities purely by:

number of jobs potentially removed.

Rank by:

value of capacity released.

This is a much more sophisticated business case.

Consider an AED 600,000 Employee

Suppose a senior enterprise salesperson costs the company:

AED 600,000 annually.

They spend 40% of their time on:

research,

CRM,

proposal coordination,

internal reporting,

follow-up administration.

That’s:

AED 240,000 worth of expensive capacity

not spent selling.

Could AI/agents remove half of that administrative burden?

Potential reclaimed capacity:

AED 120,000 equivalent.

But the real value could be much higher if the salesperson uses the time to close another:

AED 2 million

in business.

This is why AI workforce ROI should not be reduced to layoffs.

Sometimes the greatest return comes from making expensive people more economically productive.

Use AI to Increase Revenue per Employee

For growing companies, one strategic metric becomes very interesting:

revenue per employee.

Suppose UAE Company A grows revenue:

AED 50m → AED 100m

while headcount grows:

100 → 200.

Revenue per employee:

unchanged.

Company B grows:

AED 50m → AED 100m

while headcount grows:

100 → 140.

Now productivity/leverage has changed significantly.

If Company B maintains quality and sustainability, its operating model may be structurally stronger.

AI has the potential to help produce that difference.

Not by replacing all employees.

By letting the organization scale faster than its administrative headcount.

The Better Question Is “What Is Our Headcount Elasticity?”

How much does workforce size need to increase as business volume increases?

For example:

Revenue +50%.

Does headcount need:

+50%?

+30%?

+10%?

Customer volume doubles.

Does support headcount double?

Invoices double.

Does finance headcount double?

Employees double.

Does HR administration double?

AI should give some companies the ability to decouple operational volume from human headcount growth.

That’s a very powerful economic effect.

And one I would track carefully.

The Headcount Request Should Become a Capacity Request

This is perhaps the most practical organizational change.

Today:

Manager submits:

Request: Hire 3 Operations Coordinators.

Tomorrow:

Manager submits:

Capacity Requirement

Workload expected:

+40%.

Outcome required:

48-hour turnaround → 12 hours.

Current bottleneck:

document review + system coordination.

Then evaluate solutions:

Option A

3 employees.

Option B

2 employees + automation.

Option C

1 employee + agentic workflow.

Option D

Process redesign eliminates 30% of work + 1 employee.

Now leadership is solving the business problem.

Not simply approving the manager’s preferred resource.

Managers May Resist This

Understandably.

Managers know how to request employees.

They know how to manage employees.

They know what an additional person can do.

Agentic capacity feels uncertain.

It may also threaten traditional measures of managerial importance.

In some companies, larger team = more status.

If AI allows someone to run an operation with:

12 people instead of 30,

is that manager considered:

more efficient?

Or less important because their team is smaller?

Incentives matter.

A company cannot tell leaders:

“Use AI to reduce unnecessary headcount growth”

while continuing to reward managers for empire building.

Budgeting Must Change Too

Annual budgeting traditionally asks:

How many people do you need next year?

Maybe the question becomes:

What capacity do you need next year?

Then allocate across:

people,

agents,

automation,

software,

vendors.

This could create much more interesting budget discussions.

A department may request:

AED 4 million in additional capacity.

Perhaps:

AED 2 million people.

AED 1 million automation/AI.

AED 500k data/integration.

AED 500k training.

That may produce more output than:

AED 4 million payroll.

Or maybe not.

But at least leadership is comparing architectures.

HR and IT Can No Longer Plan Separately

This is another organizational consequence.

Historically:

HR plans people.

IT plans technology.

AI makes that separation increasingly strange.

Suppose HR approves:

20 additional service employees.

Meanwhile IT is building an agent expected to automate 50% of service volume.

Did anyone talk?

This sounds ridiculous.

It will happen.

Workforce planning now needs:

HR,

operations,

finance,

technology

in the same room.

McKinsey’s 2026 research argues that organizations increasingly need to rethink talent and AI resources together because agentic AI changes not simply how work is done but who or what does it. [1]

That makes talent strategy part of technology strategy.

And vice versa.

The CFO Needs to See Labor and AI on the Same Economics

Imagine:

Payroll budget:

AED 100 million.

AI budget:

AED 8 million.

The company discusses them separately.

Why?

If the AED 8 million exists partly to change the productivity of the AED 100 million, they are economically connected.

Future planning needs to consider:

workforce cost,

AI cost,

automation cost,

output,

capacity,

revenue.

Not separate technology ROI and headcount discussions.

Same operating model.

Same economics.

Build a Capacity Portfolio

A useful way to think about the future workforce is like an investment portfolio.

You need different types of capacity for different work.

Human Capacity

Best for:

ambiguity,

relationship,

leadership,

judgment,

accountability.

Automated Capacity

Best for:

clear rules,

high repetition,

structured transactions.

Agentic Capacity

Best for:

multi-step,

information-heavy,

system-connected,

bounded workflows.

External Capacity

Useful for:

specialized expertise,

variable workload,

non-core functions.

A strong company allocates intelligently across all four.

Not because one category is inherently better.

Because each has different economics and strengths.

A Decision Example

A company needs capacity to handle:

100,000 customer inquiries per year.

Option 1 — Human-Heavy

20 support employees.

Strong flexibility.

High recurring labor cost.

Option 2 — Automation-Heavy

Traditional FAQ/rules.

Cheap.

Works well on structured requests.

Weak with ambiguity.

Option 3 — Agent + Humans

AI agent resolves routine cases.

Humans handle exceptions/high-value cases.

Perhaps:

8 employees + agent infrastructure.

Potentially lower cost.

Potentially faster service.

But requires:

data,

integration,

governance,

monitoring.

Which option wins?

There is no universal answer.

That’s why management exists.

Don’t Ask AI to Justify a Bad Layoff

There is also an ethical and strategic dimension.

Leadership sometimes decides:

“Reduce headcount 20%.”

Then asks AI to make the operating model work afterward.

That’s backwards.

First redesign the work.

Measure capacity.

Understand risk.

Identify what changes.

Then make workforce decisions.

Otherwise AI becomes a story used to justify a target management already selected.

And the organization may discover later that it removed:

institutional knowledge,

customer relationships,

critical expertise,

or capacity the agents cannot actually replace.

Short-term payroll benefit can create long-term fragility.

Gartner’s Warning Is Worth Taking Seriously

Gartner’s 2026 headcount analysis argues that leaders should be cautious about simplistic AI-driven workforce-reduction assumptions.

Why?

Because AI can simultaneously:

decrease some jobs,

increase productivity,

create new roles,

increase demand elsewhere,

require AI governance,

require new technical capabilities.

Gartner even expects AI investment to produce net headcount increases in some business units through 2028, while the overall knowledge-work impact is likely to remain more mixed. [2]

That’s important.

AI transformation and workforce shrinking are not synonyms.

The Workforce Cost May Move, Not Disappear

Gartner’s June research makes another useful point:

88% of organizations it surveyed/planned around were increasing AI spending, but the technology introduces workforce costs that executives often fail to model. [3]

Examples could include:

AI engineering,

data work,

governance,

cybersecurity,

agent supervision,

reskilling.

So Company A may remove:

AED 5 million traditional labor.

Then add:

AED 2 million technology,

AED 1 million specialized talent,

AED 500k governance.

Net benefit can still be excellent.

But it isn’t:

AED 5 million.

This is why the P&L needs the complete picture.

The UAE Opportunity Is Growth Without Organizational Obesity

This may be one of the most strategically interesting implications for companies in the UAE.

The UAE continues to attract investment.

Companies are growing.

Regional expansion is strong.

AI adoption is high.

PwC’s 2026 Middle East CEO research found that 88% of CEOs planned international investment, while regional AI adoption remained significantly ahead of global averages. [4]

Growing organizations traditionally accumulate:

people,

management layers,

process,

software,

coordination.

AI offers the UAE a chance to build companies that can scale without becoming proportionally heavier.

That doesn’t mean creating companies with no employees.

It means asking:

How much organizational complexity is genuinely necessary to support growth?

That’s a much more interesting competitive advantage.

The UAE Also Needs More AI-Capable Humans

At exactly the same time, demand for AI-related talent is increasing.

PwC’s UAE AI Jobs Barometer reports that AI-related job postings more than tripled their share of UAE postings from 2021 to 2025, with growth across all examined sectors. [5]

So businesses may simultaneously:

automate some work,

avoid some hiring,

and compete more aggressively for certain people.

Again:

the workforce is not simply shrinking.

It is being recomposed.

The Human-AI Hiring Question

Before approving a new role, ask this sequence.

1. Why does this capacity need exist?

What business outcome is increasing?

2. What work is driving the requirement?

List actual activities.

3. What can disappear?

Delete before hiring.

4. What can traditional software automate?

Use the simplest reliable mechanism.

5. What can an AI agent own?

Look for:

monitoring,

coordination,

information retrieval,

classification,

multi-step execution.

6. What human capability remains?

Judgment?

Relationship?

Expertise?

Leadership?

7. Can AI multiply that human capability?

If yes, redesign the role.

8. What is the total cost of each model?

Employee.

Employee + AI.

Agent + supervisor.

Automation.

Vendor.

9. How does each model scale?

At 1× volume.

2×.

5×.

10. Which capability do we need five years from now?

Do not accidentally optimize away your future talent pipeline.

That’s a better headcount discussion.

The Hire / Augment / Agent / Automate Matrix

Use this as an executive shortcut.

HIRE

When:

human relationship is valuable,

judgment is central,

work is highly variable,

leadership/accountability matters.

HIRE + AUGMENT

When:

human expertise is valuable,

but large amounts of work involve:

research,

analysis,

drafting,

administration.

This may become one of the strongest default models.

AGENT + HUMAN SUPERVISION

When:

workflow is frequent,

bounded,

multi-step,

information-heavy,

but important exceptions require people.

AUTOMATE

When:

workflow is deterministic,

structured,

repeatable.

DELETE

When:

nobody can explain why the work exists.

Never forget the best option.

One Excellent Employee + Agents May Beat a Larger Average Team

This is where talent strategy gets interesting.

Imagine one outstanding employee can:

set direction,

manage exceptions,

judge quality,

maintain relationships,

while agents provide:

research,

coordination,

execution,

monitoring.

That person may create output previously requiring several employees.

This changes hiring.

The company may increasingly prefer:

fewer,

stronger,

more AI-capable humans

supported by machine capacity.

But there is a danger.

If you concentrate too much capability in too few people, the organization becomes fragile.

So you still need:

succession,

documentation,

development,

redundancy.

Operating leverage and organizational resilience need to be balanced.

Revenue per Employee Is Going to Become More Interesting

Especially for companies whose work can be digitally augmented.

You may see firms deliberately trying to increase:

revenue per human employee

without necessarily reducing total output or service.

Professional-services firms are already experiencing this pressure as AI changes how research, analysis, documentation and delivery are produced. [8]

That could eventually challenge entire pricing models.

If a consulting project historically required:

1,000 human hours

and now requires:

400 human hours + AI,

should the firm still sell hours?

We’ll explore business-model consequences later in the series.

The Workforce KPI I Would Add Tomorrow

Alongside:

headcount,

payroll,

turnover,

revenue per employee,

add:

Capacity Growth vs. Headcount Growth

For example:

Transaction volume:

+40%.

Revenue:

+35%.

Human headcount:

+8%.

What enabled the difference?

Automation?

Agents?

Better processes?

Improved employees?

Now you can begin measuring leverage.

This becomes far more useful than celebrating:

“AI saved 40,000 hours.”

As we argued in Article 07:

AI creates capacity. The company creates ROI.

Another KPI: New Hire Avoidance

Track:

planned hires before redesign.

Actual hires after redesign.

Adjusted for real business volume.

Example:

Historical model predicted:

15 new employees.

Actual need after AI/process redesign:

Avoided:

Now calculate legitimate cost avoidance.

This gives workforce AI projects a much clearer financial language.

Another KPI: Human Value Mix

This one is less traditional but powerful.

Estimate the share of employee time spent on:

administration,

routine execution,

judgment,

customer interaction,

innovation,

leadership.

If AI transformation works, you may want to see:

routine/admin time ↓

high-value human work ↑

That tells you whether jobs are actually changing.

Not merely whether employees have AI accounts.

Your Future Job Description May Include Agents

Imagine:

Customer Success Manager

Responsible for:

40 enterprise accounts

3 customer intelligence agents

automated renewal monitoring.

Or:

Finance Operations Manager

Responsible for:

five finance specialists

12 agents handling invoice, reconciliation and collections workflows.

Or:

Marketing Director

Responsible for:

brand/strategy team

content agents

research agents

campaign automation.

This will sound less strange very quickly.

The employee becomes responsible for outcomes produced through both human and non-human capacity.

That changes management.

Who Owns the Agent Budget?

Interesting question.

IT?

Business unit?

HR?

Finance?

If agents perform work traditionally associated with employees, they sit awkwardly between technology spending and workforce spending.

A sales department might eventually decide between:

AED 1 million additional payroll

or

AED 600,000 agent infrastructure.

Who approves that comparison?

This is another reason existing budget structures may need to change.

AI Agents Will Need Workforce Governance

Gartner predicted in April 2026 that an average global Fortune 500 enterprise could have more than 150,000 agents in use by 2028, while only 13% of organizations surveyed believed they currently had the right agent-governance approach. [9]

Whether every company reaches that exact scale is less important than the implication.

If agents proliferate, companies need to know:

which agents exist,

who owns them,

what they do,

what they cost,

which data they access,

what authority they have,

how they perform,

when they should be retired.

That looks remarkably like workforce management.

Just not HR as we know it.

Don’t Hire an Agent You Cannot Fire

Agents should have:

owners,

permissions,

performance standards,

cost limits,

lifecycle management.

If an agent:

performs badly,

becomes redundant,

uses outdated instructions,

costs more than its value,

it should be:

modified,

retrained/reconfigured,

restricted,

or retired.

Do not let agent sprawl become the next SaaS sprawl.

Companies already pay for hundreds of software subscriptions nobody remembers buying.

Imagine the same problem with autonomous software taking actions.

Not ideal.

This Is Not a Story About Fewer People

The temptation is to reduce everything to:

AI = fewer employees.

That’s too small.

The larger change is:

labor is no longer the only scalable form of cognitive capacity.

That is historically significant.

For most of modern business, if you wanted twice as much analysis, communication, coordination or administrative execution, you often needed more human labor.

AI changes the cost curve.

That allows companies to rethink:

organization size,

job design,

management,

pricing,

growth,

competitive scale.

Some will reduce headcount.

Others will grow with stable headcount.

Others will hire aggressively because AI creates new markets.

The outcome depends on strategy.

Ask “Why Hire?” Before “Who Should We Hire?”

This is the practical habit I would give every manager.

Next time a vacancy opens:

Don’t immediately replace the person.

Ask:

What outcome did the role create?

Which work still matters?

Which work can disappear?

Which work can be automated?

Which work can AI augment?

Which work requires a human?

Then write the new role.

You may discover:

the job changes.

The seniority changes.

The department changes.

Or the role disappears.

This is workforce redesign happening one vacancy at a time.

And it may be much easier than trying to redesign 5,000 jobs simultaneously.

The Vacancy Test

Every resignation is now an opportunity.

Before replacing:

1.

Do not post the old job description.

2.

Map the actual work.

3.

Delete unnecessary tasks.

4.

Automate predictable tasks.

5.

Assign appropriate tasks to agents.

6.

Design the remaining human role.

7.

Then decide whether to hire.

Imagine doing that for every vacancy for three years.

The organization would gradually redesign itself without one massive restructuring program.

That’s an interesting transformation model.

The Company After AI Will Probably Hire Differently

Not:

fewer humans everywhere.

But:

fewer humans for certain forms of coordination.

More humans with:

judgment,

expertise,

relationships,

AI fluency,

agent-management capability.

Fewer jobs defined by repetitive tasks.

More roles defined by outcomes.

More variable digital capacity.

More integration between HR and technology planning.

More scrutiny of whether headcount growth must track business growth.

That’s a much deeper workforce transformation than:

“AI will take jobs.”

The Question for Every CEO

Imagine your company grows revenue:

50%

over the next three years.

How much should headcount grow?

50%?

30%?

10%?

Maybe 80% because you’re entering a people-intensive new market.

There is no universally correct answer.

But for the first time, a CEO should be able to ask seriously:

“Why must our organization grow at the same rate as our workload?”

Somewhere inside that question is a major AI opportunity.

The Question for Every CHRO

Do you manage:

employees?

Or:

organizational capability?

If the second answer is correct, then CHRO strategy increasingly intersects with:

AI,

automation,

job redesign,

learning,

workforce architecture.

PwC’s 2026 UAE findings explicitly call for CEOs to make AI-enabled workforce transformation a leadership priority and to redesign work around human-AI collaboration instead of treating AI purely as a cost-reduction exercise. [6]

That may become one of the defining HR shifts of the decade.

The Question for Every CFO

Before approving a new permanent cost base:

Is human labor the best way to create this capacity?

Sometimes yes.

Sometimes absolutely yes.

But no longer automatically.

Compare:

human capacity,

agentic capacity,

automation,

process redesign.

Then invest.

That is capital allocation.

The Question for Every Manager

When your team says:

“We’re too busy.”

Do not immediately answer:

“I’ll request another person.”

Ask:

Why are we busy?

Because demand increased?

Excellent.

Because customers need more attention?

Important.

Because employees are performing repetitive administration?

Different problem.

Because systems don’t connect?

Different problem.

Because nobody removed an old report?

Very different problem.

Hiring should solve a human-capacity problem.

Not hide an operating-model problem.

The Question for Every Employee

There is a personal side too.

Do not only ask:

“Will AI replace my job?”

Ask:

“Which part of my work becomes more valuable when AI is available?”

If AI can handle:

research,

drafting,

coordination,

administration,

what remains?

Perhaps:

judgment,

relationships,

creative direction,

leadership,

problem framing,

accountability.

Develop those.

Because they may become the part of your job the company is increasingly willing to pay a premium for.

The Future Workforce Budget

Imagine a future department budget.

Not:

Salaries: AED 10m

Software: AED 2m

But:

Productive Capacity

Humans — AED 7m

AI/Agent Systems — AED 1.5m

Automation — AED 500k

Data/Integration — AED 500k

Specialist External Capacity — AED 500k

Total:

AED 10m.

Output:

40% greater than today’s AED 12m operating model.

Now we’re talking about transformation.

Not because AI replaced people.

Because the company learned to allocate capacity better.

So Should You Hire Another Employee?

Maybe.

If the work requires:

a person,

hire them.

Hire someone excellent.

Give them AI.

Remove the administration around them.

Let them operate at much greater leverage.

But if your next employee would spend most of their life:

copying,

searching,

formatting,

routing,

reminding,

monitoring,

updating,

coordinating,

processing predictable work,

then before opening LinkedIn Recruiter—

ask another question.

Do we need another employee?

Or do we need a better operating system?

That question did not exist in quite the same way a few years ago.

Now it belongs inside every serious workforce plan.

Because the company after AI will still have people.

Probably extremely valuable people.

But people will no longer be the automatic answer to every increase in work.

And that changes much more than hiring.

It changes:

cost,

scale,

management,

jobs,

organizational design,

and eventually—

what a company is.

What’s Next in The Company After AI

09 — Build or Buy Your AI Agents?

Once a company decides that an agentic workflow makes sense, another question arrives immediately:

Should we build it ourselves or buy an existing agent?

The wrong answer can create:

unnecessary development cost,

vendor lock-in,

weak differentiation,

integration problems,

security exposure,

and years of technical debt.

The next article will build a practical Build vs. Buy AI Agent Framework around:

strategic differentiation,

workflow complexity,

data sensitivity,

integration,

speed,

cost,

control,

vendor risk,

maintenance,

and ownership.

Because some agents should be purchased like software.

Others may become part of the company’s competitive advantage.

And those deserve to be treated very differently.

Related Udjat Insights

For businesses thinking about growth, workflow capacity and connected automation:

• Marketing Automation in Dubai — CRM + WhatsApp + Email + Sales Handoff
• How Can AI Change Marketing for Business in Dubai?
• AI Marketing Automation UAE
• Digital Transformation Agency in Dubai

Series Internal Links

As the permanent Udjat URLs become available, connect this article contextually to:

• 01 — Your Company Doesn’t Need an AI Strategy
• 02 — Stop Asking “Where Can We Use AI?”
• 04 — Don’t Automate a Bad Process
• 05 — Your First AI Agent Shouldn’t Be a Chatbot
• 06 — What Should Humans Do, and What Should AI Do?
• 07 — AI Saved 10 Hours. Where Did the Money Go?

The strongest links for Article 08 should be:

Article 04 when challenging whether the work should exist.

Article 06 when comparing human and AI decision rights.

Article 07 when calculating capacity, cost avoidance and real ROI.

Related Brightery Insights

For the systems and automation side of workforce redesign:

• Automate Company Operations
• AI HR Agent
• HR AI Empowering in Corporates
• AI-Empowered Software: Redefining Smart Business Solutions

Sources

[1] McKinsey & Company — “Rewiring Talent to Value in the Age of AI,” June 18, 2026. McKinsey describes agentic AI as changing not simply how work gets done but “who—or what” does it, creating hybrid workforces in which companies need to determine how both human talent and intelligent agents should be deployed against the highest-value work.

[2] Gartner — “Why Your Headcount Strategy Matters More Than AI Downsizing,” January 23, 2026. Gartner cautions executives against assuming AI productivity will translate directly into straightforward headcount cuts. It expects AI to reduce some roles while increasing demand elsewhere and predicts that some business units may even experience net headcount increases as AI creates new forms of work and capability requirements.

[3] Gartner — “The Hidden Workforce Costs of AI,” June 1, 2026. Gartner argues that AI is not simply reducing workforce expenditure but changing where workforce costs appear. The analysis notes rising spending expectations for AI alongside less visible costs related to implementation, skills and new workforce requirements, making traditional cost-reduction models incomplete.

[4] PwC — 29th Global CEO Survey: Middle East Findings, January 19, 2026. PwC reports strong regional AI adoption and investment, with 70% of Middle East CEOs saying they have a clearly defined AI roadmap. However, only 22% say their most-used AI tools have access to all relevant company documents and data, falling to 16% in the GCC. The report also emphasizes skills-first workforce development as AI adoption expands.

[5] PwC — 2026 AI Jobs Barometer: UAE Analysis, June 17, 2026. PwC reports that the share of UAE job postings requiring AI skills increased from 1.0% in 2021 to 3.2% in 2025, with AI-skill demand expanding across every industry studied. AI-exposed occupations are also experiencing substantially faster changes in required skills.

[6] PwC — 29th Global CEO Survey: UAE Findings, January 28, 2026. PwC reports high confidence among UAE leaders in their ability to scale AI and notes strong demand for AI talent. Its recommendations explicitly call for CEOs to redesign work around human-AI collaboration rather than treating AI primarily as a mechanism for cost reduction.

[7] Deloitte — “Future-Ready Workforce: Preparing for Work and Skills Disruption in the Agentic Enterprise,” 2025/2026 outlook. Deloitte projects that agentic AI will reshape rather than simply eliminate the workforce, shifting routine knowledge work toward automation while increasing emphasis on orchestration, oversight, exceptions and strategic decision-making. It highlights particular disruption risk for early-career work where foundational tasks are heavily automatable.

[8] McKinsey & Company — “Designing an End-to-End Technology Workforce for the AI-First Era,” April 6, 2026. McKinsey examines how organizations are redesigning technology workforces as agents take on more activity, requiring changes in hiring, internal capability development, sourcing and the relationship between technology leadership and business strategy.

[9] Gartner — “Gartner Identifies Six Steps to Manage AI Agent Sprawl,” April 28, 2026. Gartner predicts an average global Fortune 500 enterprise could have more than 150,000 AI agents in use by 2028, up from fewer than 15 in 2025, while only 13% of organizations believe they currently have the right agent-governance mechanisms. The forecast highlights the emerging management challenge created as digital labor scales.