Ask What You Would Build Differently If AI Had Always Existed

Most companies begin their AI transformation with a spreadsheet.

The departments are listed on the left.

Marketing.

Sales.

Finance.

HR.

Operations.

Customer service.

IT.

Then someone asks each department:

“Where can we use AI?”

Marketing wants AI-generated content.

Sales wants automatic follow-ups.

HR wants CV screening.

Finance wants automated reports.

Customer service wants a chatbot.

Management wants dashboards.

Everyone goes home with an AI use case.

The spreadsheet looks impressive.

The company feels innovative.

And almost nothing fundamental changes.

We have taken a company designed before generative AI, before autonomous agents, and in many cases before modern cloud software—and carefully inserted artificial intelligence into individual boxes.

That may improve productivity.

It may save time.

It may even generate a respectable return.

But we should not confuse it with transformation.

Because there is a much more important question:

If AI had existed when we designed this company, would we have designed the work this way in the first place?

That question leads somewhere completely different.

We Are Automating Decisions Made 20 Years Ago

Walk through almost any established company and you will find processes nobody intentionally designed.

They accumulated.

A customer sends an email.

Someone forwards it to another department.

That department enters the information into a spreadsheet.

Someone else checks the spreadsheet.

A manager approves something.

Another employee enters it into the CRM.

Finance receives another email.

A PDF is created.

A report is generated.

A meeting happens.

Eventually, the customer gets an answer.

Ask why the process works this way and you may hear:

“That’s how we’ve always done it.”

Then AI arrives.

And what do we do?

We add AI to step four.

The process becomes:

email → employee → spreadsheet → AI → manager → CRM → finance → PDF → meeting.

Congratulations.

We have created an AI-powered inefficient process.

This is one of the biggest risks facing companies entering the next stage of AI adoption.

AI can make bad processes faster.

And making something inefficient happen faster does not necessarily make the business better.

Digitizing a Process Is Not the Same as Redesigning It

We learned this lesson during digital transformation.

Companies frequently took paper processes and reproduced them on screens.

The form became digital.

The approval became digital.

The signature became digital.

The report became digital.

But the underlying logic remained untouched.

The company had modernized the interface without modernizing the work.

Udjat has written about this distinction in its insight on Digital Transformation Agency in Dubai: Why Smart Businesses Choose Strategy Before Technology.

Technology should follow the business problem.

Not the other way around.

AI raises the stakes because it can do something previous software could not.

Traditional software mostly required us to tell the computer exactly what to do.

AI can increasingly:

interpret,

classify,

research,

reason,

generate,

prioritize,

recommend,

communicate,

and, through agents, take actions across systems.

That means we shouldn’t merely digitize the old workflow again.

For the first time, we can reconsider why the workflow has those steps at all.

Think About Customer Service

Imagine a customer contacts a company because an order hasn’t arrived.

The traditional process may look like this:

1. Customer sends a message.

2. Customer-service employee reads it.

3. Employee identifies the customer.

4. Employee searches the CRM.

5. Employee finds the order number.

6. Employee opens the logistics system.

7. Employee checks shipment status.

8. Employee reads company policy.

9. Employee decides whether escalation is required.

10. Employee writes a response.

11. Customer receives the response.

Now ask:

Where can we use AI?

Someone will probably say:

“Step 10. AI can write the email.”

Great.

We saved perhaps 60 seconds.

Now ask the better question:

If we designed this process today, what would it look like?

The system could identify the customer automatically.

Retrieve the order.

Check delivery status.

Compare the situation against company policies.

Determine whether it falls within an approved resolution threshold.

Initiate a replacement or refund when permitted.

Update the CRM.

Generate the communication.

Escalate only exceptions requiring human judgment.

Suddenly the question isn’t:

How can AI make the employee write faster?

It is:

Why does the employee need to manually coordinate this process at all?

That is workflow redesign.

The Unit of AI Transformation Is Not the Task

This is a subtle but important difference.

Most AI adoption focuses on tasks.

Write this.

Summarize that.

Analyze this spreadsheet.

Generate those images.

Research this company.

Draft this email.

Tasks are easy to see.

But businesses don’t produce value through isolated tasks.

They produce value through workflows.

A sales workflow might include:

lead capture →

data enrichment →

qualification →

assignment →

research →

outreach →

follow-up →

meeting →

proposal →

approval →

contract →

onboarding.

Automating one task inside that sequence can be helpful.

But redesigning the sequence itself can change the economics of the entire function.

This is why Brightery’s work around automating company operations emphasizes structured workflows rather than simply introducing more tools.

The opportunity is not:

AI does one task faster.

The bigger opportunity is:

The business requires fewer tasks to achieve the same outcome.

That distinction matters enormously.

Don’t Ask What AI Can Do

There is another problem with starting from AI capabilities.

AI changes too quickly.

If your strategy begins with:

“Here are the things the technology can currently do.”

your transformation program immediately starts aging.

A better starting point is stable:

What outcome are we trying to produce?

Customers need their problems resolved.

Sales needs qualified opportunities converted.

Finance needs accurate financial control.

Operations needs products or services delivered.

HR needs the right people deployed effectively.

Management needs decisions made.

The outcome changes much more slowly than the technology.

So begin there.

Then redesign backward.

Start With the Outcome

Imagine your sales team wants to improve qualified meeting conversion.

The traditional conversation might be:

“Which AI sales tool should we buy?”

The redesigned conversation starts differently.

Desired outcome

Increase qualified sales conversations without proportionally increasing sales headcount.

Now map everything that has to happen:

Leads arrive.

Information is collected.

Duplicates are removed.

Companies are researched.

Intent is assessed.

Prospects are scored.

Accounts are prioritized.

Messages are personalized.

Follow-ups happen.

Responses are categorized.

CRM records are updated.

Meetings are booked.

Salespeople prepare.

Now you can ask a much more useful set of questions.

Which work requires human judgment?

Which work requires human relationships?

Which work is repetitive?

Which work requires searching information?

Which decisions follow predictable rules?

Which work could an AI agent perform?

Which actions can be executed automatically?

Where does a human need to approve?

Where do exceptions occur?

What should disappear entirely?

The result may look nothing like the existing sales process.

And that’s exactly the point.

Four Ways AI Can Change Work

I think leadership teams need a simple vocabulary for this.

Every workflow should be evaluated through four possibilities.

1. Assist

The human still performs the work.

AI helps.

Examples:

drafting,

research,

analysis,

summarization,

translation,

brainstorming.

This is where much enterprise AI adoption currently sits.

Useful.

Low disruption.

Usually easy to implement.

But limited structural transformation.

2. Automate

AI or software executes a defined piece of work.

Examples:

classifying inquiries,

extracting invoice data,

generating routine reports,

routing requests,

updating records,

sending standard follow-ups.

The human performs less of the workflow.

Efficiency can increase substantially.

But we are still working largely within the existing process.

3. Orchestrate

This is where agents become more interesting.

Instead of automating one step, an AI system coordinates multiple steps.

It may:

receive a request,

retrieve information,

use several systems,

apply company policies,

make bounded decisions,

trigger actions,

communicate results,

and escalate exceptions.

The employee moves from performing every step to supervising the outcome.

This is a fundamentally different operating model.

4. Eliminate

This is the category executives should pay more attention to.

Sometimes AI doesn’t improve a step.

It removes the reason the step exists.

Consider internal reporting.

A company may employ analysts to:

extract data,

combine spreadsheets,

prepare slides,

write summaries,

email reports,

and then explain those reports to management.

You could use AI to generate the slides faster.

Or you could ask:

Why is management waiting for a monthly PowerPoint at all?

What if leaders could query validated business information directly?

What if exceptions were surfaced continuously?

What if important deviations triggered analysis automatically?

What if the “report” as we understand it stopped existing?

Then the transformation isn’t:

faster report production.

It is:

less need for reports.

That’s where AI begins changing organizations rather than tasks.

The Most Important Button Might Be Delete

There is an instinct inside technology projects to preserve existing complexity.

Every requirement becomes sacred.

“This approval must stay.”

“This report must stay.”

“This step must stay.”

“This spreadsheet must stay.”

“This system must stay.”

Then the transformation team is asked to connect everything.

Sometimes that’s necessary.

But every AI redesign project should have a question nobody likes asking:

What can we delete?

Delete the report.

Delete the handoff.

Delete the duplicated data entry.

Delete the unnecessary approval.

Delete the meeting.

Delete the manual classification.

Delete the request for information the system already possesses.

Delete the software screen nobody actually needs.

Only after elimination should we decide what needs automation.

This matters because complexity has a cost.

Every additional step introduces:

time,

errors,

handoffs,

training,

software,

management,

exceptions,

and opportunities for failure.

AI gives companies an opportunity to automate.

But its more valuable gift may be forcing us to question why so much work exists.

An AI Workflow Should Be Designed Around Exceptions

Traditional processes often require humans to touch every transaction.

Consider invoice processing.

Suppose 10,000 invoices arrive.

Perhaps 8,500 are completely ordinary.

Purchase order matches.

Supplier matches.

Amounts match.

Required information exists.

Nothing unusual happened.

Why does a human need to inspect all 8,500?

Historically the answer might have been:

because software wasn’t capable enough.

But suppose AI and automation can reliably process normal transactions.

Then the workflow changes.

Humans don’t manage transactions.

They manage exceptions.

The system handles:

normal.

normal.

normal.

normal.

exception.

The human receives:

“This invoice is 18% above the purchase-order value and doesn’t match the agreed supplier terms.”

Now human attention is concentrated where judgment actually creates value.

That is an important principle for the company after AI:

Do not distribute human attention equally across work that does not require equal judgment.

Human attention is expensive.

Spend it accordingly.

This Changes the Role of the Employee

There is an uncomfortable consequence.

When workflows change, jobs change.

Consider an employee whose day currently consists of:

30% collecting information,

20% preparing documentation,

15% entering information,

10% checking status,

10% routine communication,

10% judgment,

5% relationships.

An AI project may automate individual pieces.

But an AI-native redesign asks:

If the first 70% disappears or becomes agent-operated, what is this person’s job now?

Maybe the employee handles:

exceptions,

judgment,

customer relationships,

quality control,

process improvement,

agent supervision,

and difficult decisions.

That can create a much better job.

But it isn’t the same job.

This is why AI transformation eventually stops being an IT initiative.

It becomes:

organizational design,

workforce strategy,

management design,

training,

compensation,

governance,

and culture.

We will explore these consequences later in The Company After AI, particularly when we ask whether companies should hire an employee or deploy an agent—and what happens to middle management.

Human in the Loop Is Not One Thing

Companies also need to become more precise about “human oversight.”

People often say:

“Don’t worry. There will always be a human in the loop.”

Fine.

Where?

Before the AI acts?

After it acts?

Only above a certain financial value?

Only for unusual cases?

Random quality control?

Final approval?

Policy creation?

Escalation?

There are very different ways humans can participate.

Human Does the Work

AI provides support.

Human Approves the Work

AI performs it but cannot act without permission.

Human Monitors the Work

AI acts within predefined boundaries while humans supervise.

Human Handles Exceptions

AI completes normal work and escalates unusual cases.

Human Governs the System

AI operates largely autonomously while humans define policies, limits and accountability.

Each model creates different:

cost,

speed,

risk,

staffing,

technology,

and governance requirements.

“Human in the loop” is therefore not a strategy.

It is a design decision.

The Company Should Be Designed Around Outcomes, Not Departments

Here is where things become more interesting.

Traditional organizations are divided into functions.

Marketing.

Sales.

Finance.

Operations.

HR.

Technology.

But customers don’t experience departments.

A customer experiences an outcome.

Imagine someone buys a product.

That journey may touch:

marketing,

e-commerce,

payments,

inventory,

warehouse,

logistics,

customer service,

finance,

and CRM.

Every departmental boundary becomes a potential handoff.

AI agents don’t necessarily care about those boundaries.

An agent can potentially operate across systems and coordinate work spanning several functions—as long as permissions, data, architecture and governance allow it.

This raises a significant organizational question:

Should we continue designing companies around departments—or increasingly around outcomes?

For example:

Not:

Marketing department + Sales department + Customer service department

But perhaps:

Customer acquisition

Customer conversion

Customer success

with humans and agents organized around each outcome.

This won’t happen everywhere.

And it won’t happen overnight.

But AI makes the question possible in a way traditional automation didn’t.

McKinsey describes an emerging version of this as the agentic organization, where human and AI teams increasingly organize around outcomes rather than traditional functional silos. [1]

That makes the operating model—not merely the AI model—one of the most important competitive questions ahead.

Before You Automate Anything, Draw the Workflow

This is where I’d tell companies to become surprisingly low-tech.

Forget AI for an hour.

Get the people who actually perform the work into a room.

Put the workflow on a wall.

Start with:

Trigger

What starts the process?

Outcome

What must be true when it finishes?

Steps

What happens between them?

Decisions

Who decides what?

Information

What information is required?

Systems

Where does the work happen?

Handoffs

Where does responsibility move between people or departments?

Waiting

Where does nothing happen?

Exceptions

What causes the process to break?

Controls

Which checks exist because of regulation, risk or quality?

Customer impact

Where is the customer waiting or doing unnecessary work?

Only then introduce AI.

The AI Redesign Test

For every step, ask seven questions.

Question 1

Why does this step exist?

This sounds obvious.

It isn’t.

If nobody can explain why, that’s useful information.

Question 2

Can we eliminate it?

Deletion comes first.

Question 3

Can a rule handle it?

Do not use artificial intelligence when deterministic automation is simpler, cheaper and safer.

An “AI transformation” doesn’t require AI in every step.

Sometimes:

if amount < AED 1,000 → approve

is better than asking a language model to think about it.

Question 4

Can AI assist the human?

Use AI when language, interpretation, synthesis or prediction improves the person’s work.

Question 5

Can AI perform it with human approval?

Useful when the task is predictable but consequences matter.

Question 6

Can an agent execute it autonomously within limits?

Define:

permissions,

financial thresholds,

allowed systems,

actions,

exceptions,

logging,

and escalation rules.

Question 7

What should humans own?

Do not answer this last.

Start designing the human role intentionally.

Humans should not merely receive whatever tasks are left after automation.

Decide where human capabilities create disproportionate value:

judgment,

responsibility,

relationships,

negotiation,

creativity,

context,

ethics,

leadership,

taste,

and decisions under genuine ambiguity.

A Simple Example: Lead Management

Let’s redesign something familiar to many UAE businesses.

An inquiry arrives through the website.

The old process

Form submitted.

Email notification sent.

Marketing downloads leads.

Spreadsheet updated.

Sales manager reviews.

Lead assigned.

Salesperson researches company.

Salesperson checks previous activity.

Salesperson writes WhatsApp/email.

Salesperson follows up.

Salesperson updates CRM.

Manager reviews pipeline.

Report prepared.

Meeting held.

Familiar?

Now imagine asking:

Where can AI help?

You may automate the email.

You may summarize the lead.

You may generate the follow-up.

That’s useful.

Udjat’s work on AI Marketing Automation UAE already explores how AI can improve lead qualification, segmentation, customer journeys and marketing-to-sales handoffs.

But redesign the workflow from zero.

The redesigned process

Lead enters.

↓

Identity and company data enriched automatically.

↓

CRM history retrieved.

↓

Source, behavior and intent analyzed.

↓

Lead categorized against qualification criteria.

↓

High-potential opportunities assigned immediately.

↓

Account research assembled automatically.

↓

Recommended sales approach produced.

↓

Appropriate first communication prepared or sent according to policy.

↓

Response monitored.

↓

Follow-up coordinated.

↓

CRM maintained automatically.

↓

Salesperson receives the opportunity when human interaction creates value.

Now the salesperson’s job changes.

Less:

copying,

researching,

typing,

updating,

checking.

More:

understanding,

negotiating,

building trust,

discovering needs,

designing solutions,

closing.

That is what AI transformation should look like when it works well.

The Goal Is Not Fewer Humans

It may produce fewer jobs in some workflows.

We should not pretend otherwise.

It may also create more capacity.

It may let smaller companies compete with much larger organizations.

It may move employees toward more valuable work.

It may create entirely new services.

It may make previously uneconomic customer experiences affordable.

But “reduce headcount” is too narrow to be the design objective.

The objective should be:

Produce the desired business outcome using the best combination of humans, AI, automation and software.

Sometimes the answer requires more humans.

Sometimes fewer.

Sometimes different skills.

Sometimes an agent.

Sometimes simple software.

Sometimes nothing needs changing.

The correct answer is determined by the outcome.

Not the technology.

The UAE Opportunity Is Larger Than Cost Reduction

There is a particular danger for companies thinking about AI only through efficiency.

The UAE is not a low-ambition market.

Businesses here compete on:

speed,

service,

experience,

personalization,

international reach,

convenience,

and increasingly digital sophistication.

So AI workflow redesign should not only ask:

“How can we do this with fewer resources?”

It should also ask:

“What could we offer that wasn’t possible before?”

Could every customer receive a personalized experience?

Could service become effectively 24/7 without becoming robotic?

Could a small professional-services company deliver the research capacity of a much larger firm?

Could a real-estate business respond intelligently to thousands of inquiries instead of simply collecting them?

Could hotels understand and respond to guest preferences before arrival?

Could retailers personalize interactions across languages and channels?

Could professional firms turn decades of internal knowledge into something every employee can access?

Could managers receive exceptions and decisions rather than waiting for reports?

The most interesting AI transformations won’t simply make the old company cheaper.

They will make different companies possible.

This Is Where AI Agents Matter

Much of the first wave of generative AI focused on creation.

Write this.

Generate that.

Summarize this.

Agents change the conversation because they are increasingly designed to perform multi-step work.

They can potentially:

understand an objective,

determine necessary steps,

retrieve information,

use software,

coordinate with other agents,

execute actions,

evaluate results,

and escalate when required.

That changes the unit of automation.

We move from:

task

to:

workflow

and eventually perhaps:

outcome.

Deloitte’s 2026 research found that while only 6% of surveyed leaders said more than 40% of their organizational processes were currently automated or AI-enabled, 42% expected that level by 2028. [2]

Whether that expectation proves completely accurate is less important than what it tells us about executive direction.

Companies are preparing for AI to enter operational workflows at scale.

And if that happens, simply attaching agents to workflows designed entirely for humans is unlikely to be enough.

AI-First Does Not Mean AI Everywhere

This deserves emphasis.

Designing AI-first is not the same thing as inserting AI into everything.

AI-first means:

We acknowledge the capability exists before we design the workflow.

Then we choose the best architecture.

A process might contain:

traditional software,

deterministic automation,

AI models,

agents,

human employees,

external partners,

APIs,

and manual controls.

Good architecture chooses the simplest reliable mechanism for each job.

For example:

A calculator should calculate.

A database should store structured data.

A rules engine should enforce clear rules.

An AI model can interpret ambiguity.

An agent can coordinate actions.

A human should make decisions where judgment, accountability or trust matters.

The goal is not maximum AI.

The goal is minimum unnecessary work.

What Would You Build If You Started Today?

This may be the most useful workshop exercise a leadership team can perform.

Pick one critical business outcome.

Not a department.

An outcome.

For example:

Convert a qualified opportunity into a customer.

Then imagine you are launching the company tomorrow.

There are no legacy systems.

No existing departments.

No old job descriptions.

No historical procedures.

No one can say:

“But this is how we do it.”

You have:

modern software,

APIs,

enterprise data,

generative AI,

AI agents,

automation,

and humans.

Now ask:

What is the shortest reliable path from trigger to outcome?

What information is needed?

Which decisions are deterministic?

Which decisions require judgment?

What can happen automatically?

What requires human interaction?

What should trigger an exception?

Where should accountability sit?

What should the customer experience?

What does success cost?

Then compare your new workflow with the current one.

The gap between them is not simply an automation opportunity.

It is your transformation opportunity.

The 5-Step AI Workflow Redesign Framework

For leadership teams wanting something practical, I would use this sequence.

Step 1 — Choose an Outcome

Bad:

“Automate finance.”

Better:

“Reduce invoice-to-payment cycle time while maintaining financial controls.”

Bad:

“Use AI in customer service.”

Better:

“Resolve 80% of routine customer issues immediately without reducing satisfaction.”

Make the outcome measurable.

Step 2 — Map the Current Reality

Not the official SOP.

The actual process.

Include:

manual work,

spreadsheets,

WhatsApp,

email,

shadow systems,

phone calls,

approval habits,

workarounds,

and the spreadsheet Ahmed built in 2019 that apparently runs half the company.

Every organization has one.

Step 3 — Delete Before Automating

For every activity:

Why?

What happens if we remove it?

Does regulation require it?

Does the customer value it?

Does another system already perform it?

Can two steps become one?

This is often where the cheapest productivity gains appear.

No AI required.

Step 4 — Allocate the Work

For everything remaining, choose:

Human

Human + AI

AI + approval

Automated system

AI agent

Agent + exception handling

Do not allocate work based purely on technical possibility.

Allocate based on:

value,

cost,

risk,

quality,

speed,

judgment,

trust,

and accountability.

Step 5 — Redesign the Organization Around It

This final step is frequently forgotten.

If the workflow changes, ask:

Does the role change?

Does the manager’s responsibility change?

Does the KPI change?

Does the approval structure change?

Does compensation change?

Do permissions change?

Does training change?

Does the department boundary still make sense?

Otherwise the old organization will eventually force the new workflow back into the old way of working.

The Transformation Paradox

Microsoft’s 2026 Work Trend Index describes what it calls the Transformation Paradox.

Employees are becoming capable of using increasingly sophisticated AI, while the organizations around them often remain designed for older ways of working. Microsoft found that organizational factors such as culture, management support and talent practices account for roughly twice the reported impact of individual factors. [3]

There is another finding worth noticing.

Only 19% of AI users in Microsoft’s global research were in what it describes as the “Frontier” position, where both individual AI capability and organizational readiness were high. [3]

That tells us something important.

The bottleneck is increasingly not:

Can employees use AI?

It is:

Can the organization absorb what AI makes possible?

An employee may be able to produce an analysis in 15 minutes.

Then wait three days for approval.

A salesperson may understand the prospect immediately.

Then spend 20 minutes entering CRM data.

AI may generate a decision recommendation instantly.

Then six people attend a meeting to discuss it.

The employee became faster.

The organization did not.

That is exactly why workflow redesign matters.

The Competitive Advantage May Not Be Better AI

Companies are naturally obsessed with models.

Which model is smartest?

Which benchmark is highest?

Which provider is winning?

Those questions matter to technologists.

They may matter less strategically than businesses expect.

Because competitors increasingly have access to many of the same foundation models.

Your competitor can access advanced AI too.

They can buy similar copilots.

Similar agent platforms.

Similar APIs.

Similar models.

So where does advantage come from?

Potentially from:

your proprietary data,

your customer relationships,

your institutional knowledge,

your product,

your decisions,

your integrations,

your talent,

and—critically—

the way your company is designed to use AI.

Two companies can have access to the same intelligence and produce completely different outcomes.

The difference is the operating system around it.

McKinsey’s July 2026 analysis reaches essentially this conclusion: many companies have deployed AI tools, but relatively few are seeing enterprise-wide financial impact because they are accelerating existing activities while leaving governance, teams, capabilities and the underlying operating model largely unchanged. [4]

That is the new competitive battlefield.

Not merely:

Who has AI?

Almost everyone will.

But:

Who redesigned the business around what AI makes possible?

Before Your Next AI Investment

Before purchasing another AI platform, ask:

Which business outcome are we changing?

What is the current workflow?

Why does every step exist?

Which steps can disappear?

Where does human judgment genuinely matter?

Which work can traditional automation handle?

Where does AI add something fundamentally different?

Where can an agent own multiple steps?

What systems must connect?

What data is required?

What permissions are necessary?

What happens when the AI is wrong?

What metric should improve?

And perhaps most importantly:

If we were designing this process today, would we build anything resembling what we have now?

If the answer is no, don’t automate it yet.

Redesign it.

The Company After AI Is Not the Old Company + AI

That may become the central idea of this entire series.

The company after AI isn’t necessarily:

your company

ChatGPT

Copilot

a chatbot

several agents.

Just as the internet eventually changed far more than corporate communications, AI may eventually change far more than employee productivity.

It can change:

how work moves,

where decisions happen,

what jobs contain,

what managers manage,

how companies scale,

how customers are served,

how software is designed,

and what an organization economically needs to look like.

But that value will remain hidden if we continue asking:

“Where can we insert AI into the company we already have?”

There is a better question.

What company would we build now?

And once you ask that question seriously, transformation begins to mean something very different.

What’s Next in The Company After AI

03 — Your AI Pilot Worked. Your Transformation Didn’t.

Companies around the world have successful AI demonstrations.

The chatbot works.

The assistant works.

The prototype works.

Employees like it.

Leadership approves the next pilot.

And yet enterprise-wide value remains surprisingly difficult to find.

Why?

The next article examines the gap between AI experimentation and actual business transformation—and why some organizations become permanently trapped in pilot mode.

Related Udjat Insights

For readers working on the strategy and execution behind this transition:

• Digital Transformation Agency in Dubai: Why Smart Businesses Choose Strategy Before Technology
• AI Marketing Automation UAE
• How Can AI Change Marketing for Business in Dubai?

Series internal link: Once Article 01 is live, link the first reference to this series back to “Your Company Doesn’t Need an AI Strategy” using its final Udjat URL.

Related Brightery Insights

For the technology, automation and implementation side:

• Automate Company Operations
• AI Automation in Software
• AI-Empowered Software: Redefining Smart Business Solutions

Sources

[1] McKinsey & Company — “The agentic organization: Contours of the next paradigm for the AI era,” September 26, 2025. McKinsey describes a movement toward AI-first workflows and human-agent teams organized increasingly around outcomes rather than traditional functional silos.

[2] Deloitte — “Rewiring the enterprise operating model for AI scale,” July 2026. Based on Deloitte’s 2026 Global Technology Leadership Study of more than 660 technology executives. Nearly three-quarters of surveyed executives said their operating model would need to change over the following 12–18 months; 42% expected more than 40% of organizational processes to be automated or AI-enabled by 2028, compared with 6% at the time of the survey.

[3] Microsoft — “2026 Work Trend Index Annual Report: Agents, Human Agency, and the Opportunity for Every Organization,” May 5, 2026. Microsoft analyzed Microsoft 365 productivity signals and surveyed 20,000 AI-using workers across 10 countries. Its research describes a “Transformation Paradox” in which worker capability can move faster than organizational readiness; only 19% of surveyed AI users fell into its high-capability/high-readiness “Frontier” group.

[4] McKinsey & Company — “The operating model advantage: Why AI winners are rewiring their organizations,” July 7, 2026. McKinsey argues that many companies are using AI to accelerate existing activities while leaving their underlying governance, team structures and operating models largely unchanged, limiting enterprise-wide financial impact.

[5] Deloitte — “The agentic reality check: Preparing for a silicon-based workforce,” Tech Trends 2026. Deloitte argues that businesses frequently attempt to automate processes originally designed for human workers rather than redesigning end-to-end operations for agentic AI.

[6] Gartner — “How to Reimagine Any Process and Team for AI,” July 1, 2026. Gartner’s research frames AI adoption as an opportunity to redesign both process execution and outcomes rather than limiting adoption to individual tools or tasks.