AI Can Now Work Across Departments. Your Company Is Still Organised Around Them. 

AI is beginning to do more than answer questions. It can execute tasks, coordinate workflows, interact with customers, and increasingly take on work that once moved between people and departments. The question for businesses is no longer simply how to adopt AI, but what needs to be redesigned when AI becomes part of how work gets done.

Firuz Alimov has spent more than two decades working across finance, banking, telecommunications, government, consulting, academia, startups, and emerging technology.

Today, he is Country Director Malaysia at The AI Collective, Founder of venture studio Aether Builders, a management consultant at Alimov Ltd, and a Deal Partner and Boardy Pro Advisory Board member at Boardy. His work spans AI ecosystems, venture building, operating models, fundraising, and business transformation — giving him a view of AI not simply as a technology to deploy, but as something that changes how companies operate.

In this conversation with NewInAsia, Alimov examines what SMEs should fix before introducing AI, why AI agents turn a technology decision into an organizational one, and what Malaysian businesses risk getting wrong as adoption accelerates.

Before You Buy Another AI Tool, Fix This First

Q: You recently argued that SMEs don’t need “AI magic” — they need AI that works. What should a business fix before introducing AI, rather than expecting AI to fix it?

A: The biggest misconception I see is that AI is a solution looking for a problem.

A business owner attends an AI conference, sees a few impressive demos, hears someone claim they increased productivity by 300 percent, and immediately decides the company needs AI. The following Monday, they call the management team and announce that AI will transform everything.

Three months later, nothing has changed except the number of subscriptions being paid every month.

The uncomfortable truth is that AI amplifies whatever already exists inside the business.

When your processes are broken, AI simply helps you break them faster. Similarly, if your data is messy, AI will create more confusion. Finally, where nobody knows who owns a process, AI just gives everyone a new thing to blame.

Before introducing AI, companies need to answer a very boring question: How does work actually get done today?

Most SMEs cannot answer that clearly.

Ask five people how a customer enquiry becomes a sale, and you will receive seven different answers. Ask who owns follow-up, approvals, customer onboarding, invoicing, collections, and reporting, and suddenly everyone becomes interested in the weather.

The companies seeing results from AI are not necessarily the most advanced. They are the ones that understand their workflows, decision points, bottlenecks, and accountability structures.

AI is not magic. It is a multiplier.

And multipliers are dangerous because they make good systems better and bad systems worse.

The businesses winning with AI are not starting with ChatGPT. They are starting with process mapping, operational clarity, and identifying repetitive work that nobody enjoys doing in the first place.

The irony is that the less glamorous work usually creates the biggest AI returns.

Nobody posts LinkedIn updates celebrating automated invoice reconciliation. Yet that boring automation often creates more value than a thousand AI-generated social media posts.

The future belongs to companies that stop asking, “How do we use AI?”

And start asking, “Where is work getting stuck?”

The Moment AI Becomes a Management Problem

Q: We are moving from AI that answers questions to agents that can execute tasks and pursue goals. At what point does adopting AI stop being a technology decision and become an organizational design decision?

A: I would argue we have already crossed that line.

When AI was simply answering questions, it was a software discussion. When AI starts executing tasks, making recommendations, coordinating workflows, and interacting with customers, it becomes an organizational discussion.

Imagine hiring twenty highly capable interns tomorrow.

Would your biggest challenge be the technology?

Of course not.

Your biggest challenge would be defining responsibilities, approvals, reporting lines, governance, performance expectations, and accountability.

Agentic AI introduces exactly the same challenge.

Many leaders still think of AI as software. Increasingly, they should think of AI as digital labour.

The question is no longer whether AI can perform work. The question is who owns the outcome when that work is performed.

This is where many organizations become uncomfortable.

Traditional companies are built around functions. Marketing reports to marketing. Finance reports to finance. Operations reports to operations.

AI does not particularly care about organizational charts.

An AI agent processing customer onboarding may touch sales, compliance, finance, operations, and customer support within a single workflow. Suddenly, the biggest challenge is not capability. It is coordination.

For decades, organizations optimized around information scarcity. People existed partly because information had to travel through them.

Today, information moves instantly. Tomorrow, decisions will increasingly move instantly as well.

That forces leaders to redesign how authority, accountability, and execution interact.

The companies that thrive will not simply deploy AI agents. They will redesign the company around a world where agents exist.

The companies that struggle will attempt to force twenty-first-century technology into twentieth-century operating models.

That usually ends about as well as installing a Formula One engine into a horse cart.

What If the Org Chart Is the Legacy System?

Q: If an AI agent can eventually perform work across departments, should companies still organize themselves around traditional functions — or does agentic AI make some of today’s organizational structures obsolete?

A: This might sound controversial, but many organizational structures were already becoming obsolete before AI arrived.

Most modern companies still operate using a model that would be familiar to military commanders thousands of years ago.

You have layers of hierarchy. Information flows upward. Decisions flow downward. Each department owns a specific function. Finance handles finance. Marketing handles marketing. Operations handles operations.

Everyone attends meetings to explain to everyone else why something has not happened yet.

That structure made sense when communication was slow, information was scarce, and coordination was expensive.

The problem is that we no longer live in that world.

The trend has already been visible for years. Many of the world’s fastest-growing companies operate with remarkably small teams compared with previous generations.

A startup with twenty people can now achieve what once required two hundred. A company with two hundred can increasingly compete against organizations with thousands.

This is not because people suddenly became smarter. It is because technology reduced coordination costs.

AI accelerates that trend dramatically.

Traditional departments exist largely because work needs to be divided, managed, coordinated, reported, approved, and handed off between specialists. Agentic AI challenges many of those assumptions.

An AI-powered workflow does not care whether a task belongs to sales, customer service, operations, finance, or compliance. It simply executes the process.

What matters is the outcome, not which department owns the box on the organizational chart.

I suspect the future organization will look less like a pyramid and more like a collection of small, highly accountable teams supported by large numbers of AI agents.

Humans will increasingly focus on strategy, relationships, judgment, creativity, negotiation, and exception handling. AI will handle much of the coordination, administration, reporting, information gathering, and execution.

The result is not necessarily fewer companies. It is companies that move faster.

Organizations will require fewer management layers, adapting more quickly to market changes. As a result, companies will spend less time managing bureaucracy and more time serving customers.

This does not mean organizational structure disappears.

It means organizational structure becomes flatter, more fluid, and more outcome-driven.

For decades, businesses optimized around managing people. Increasingly, they will optimize around managing systems.

The companies that thrive will not ask, “How do we fit AI into our existing departments?”

They will ask a much harder question: “If we were building this company from scratch today, would we organize it this way at all?”

That question makes many executives uncomfortable.

It is also the question AI forces every leadership team to confront.

Maybe Employees Aren’t Resisting AI — Maybe They’re Resisting Bad Leadership

Q: You often emphasize incentives, ownership, and human behaviour. When AI implementation fails, how much of the resistance attributed to “people fearing AI” is actually rational resistance to poorly designed change?

A: Far more than most executives want to admit.

Blaming employees is often the easiest explanation.

“It failed because people resisted AI.”

That statement sounds neat and convenient. It is also frequently wrong.

Imagine your boss arrives tomorrow and announces that an AI system will evaluate performance, recommend promotions, monitor productivity, and potentially influence staffing decisions.

Then imagine nobody explains how any of it works.

Would you be excited? Or would you be concerned?

Most people are not resisting AI. Instead, employees are resisting uncertainty, poor communication, and incentives that suddenly appear misaligned.

In many organizations, employees possess legitimate concerns.

If AI saves ten hours per week, who benefits? The company? The customer? Or the employee who created those efficiencies?

If AI changes how decisions are made, who remains accountable when mistakes occur?

These are not emotional questions. They are rational questions.

Every major technological shift in history created similar tensions. The difference today is the speed.

The conversation often becomes framed as humans versus AI. I think that is the wrong framing entirely.

The real question is whether leaders are designing systems where humans and AI can work together effectively.

When implementation succeeds, employees typically become enthusiastic very quickly because nobody enjoys repetitive administrative work.

When implementation fails, it is usually because leaders focused on the technology while ignoring trust, incentives, communication, and organizational psychology.

Technology changes systems.

People determine whether systems survive.

The Bigger Risk for Malaysia Isn’t Moving Too Slowly

Q: Malaysia wants to accelerate practical AI adoption, particularly among SMEs. What would concern you more over the next few years: Malaysian businesses adopting AI too slowly, or adopting it quickly without redesigning how decisions, accountability, and work actually happen?

A: Without question, the second scenario.

Slow adoption can be fixed.

Bad adoption creates expensive lessons.

Malaysia is entering a fascinating phase. The conversation is shifting from AI awareness to AI implementation. That is positive.

The danger is that companies mistake activity for progress.

Buying AI tools is not transformation.

Running pilot projects is not transformation.

Attending workshops and posting photos on LinkedIn is definitely not transformation.

Transformation occurs when the economics of the business change.

Transformation happens when decisions become faster, customer experiences improve, costs decrease, revenue grows, and new capabilities emerge.

What worries me is that many organizations are focusing on technology acquisition while neglecting organizational redesign.

AI is not another software upgrade.

AI transforms how information flows, reshaping how decisions are made and who performs the work. And eventually, it changes the number of people required for certain activities.

Those realities require leadership courage.

The most dangerous outcome is creating what I call “AI theatre.”

AI theatre creates a flurry of announcements, dashboards, and pilot programs.

Very little measurable impact.

Malaysia has an opportunity to leapfrog because SMEs are generally more agile than large enterprises. They have fewer layers, less bureaucracy, and can move faster when leadership is committed.

But speed without clarity creates chaos.

My concern is not that Malaysian businesses will adopt AI.

My concern is that some will automate yesterday’s operating model instead of redesigning tomorrow’s.

The winners will not be the companies with the most AI.

They will be the companies that rethink how work itself gets done.

That is a much harder conversation.

It is also where the real value is created.

Final Thoughts: The Question Leaders Can’t Avoid Much Longer

As AI takes on more execution, the challenge for companies will not simply be deciding which model, platform, or agent to deploy.

It will be deciding what has to change around it.

Workflows designed around human hand-offs may no longer make sense. Departments created around information bottlenecks may need to become more fluid. Accountability systems will have to evolve when work is shared between people and AI.

For SMEs, that could become an advantage. Smaller organizations often have fewer layers to redesign and less bureaucracy standing in the way.

But adopting more AI is not the same as becoming an AI-enabled organization.

The deeper leadership question is the one running through Alimov’s argument:

If AI can change how work gets done, what would you build differently if you were starting the company today?

Read the Chinese article here.