AI Has Moved Beyond the Pilot. Now the Operating Model Has to Catch Up.

Claritiv Technologies

4 days ago

5 min read

Artificial intelligence has moved rapidly from experimentation into everyday business use.

Employees are using AI to analyse information, write content, support customer interactions, generate code, summarise documents and automate repetitive work. At the same time, organisations are investing heavily in AI infrastructure, platforms and applications.

The question facing business leaders in late 2026 is therefore changing.

The issue is no longer whether AI has a place in the organisation. For many businesses, it already does. The harder question is whether the organisation itself has changed enough to capture the value that AI makes possible.

That distinction matters.

According to McKinsey's 2026 State of AI research, nearly nine in ten respondents say their organisations regularly use AI in at least one business function. At the same time, 44% report that AI is scaling across their enterprise, up from 38% a year earlier. AI is therefore moving beyond isolated experimentation and becoming part of a broader operating environment.

Yet adoption alone does not guarantee business value.

Gartner's October 2026 research makes the issue particularly clear: organisations struggling to achieve enterprise value from AI need to rethink how work and decisions operate around AI rather than simply adding AI to existing workflows.

For business leaders, this creates a very different management challenge.

AI changes the way work gets done

Every technology eventually becomes part of a wider operating model. AI is reaching that point quickly.

Consider a customer service team. An AI assistant might summarise customer interactions, recommend responses and retrieve information from internal systems. The technology may work perfectly. But what happens next?

Who reviews the recommendation? When can the system act without approval? What happens when the information is incomplete? Which customer interactions should be escalated? How are decisions recorded? How does the organisation learn from the outcomes?

These are operating questions.

The same applies to software development. AI coding tools can accelerate development, generate code and help teams work through technical problems. But if product priorities remain unclear, requirements continue changing late in the process, architecture is poorly understood and teams lack appropriate review processes, faster coding does not automatically produce better products.

AI can increase the speed of work while leaving the underlying system of work unchanged.

That is where many organisations are now finding the limits of AI experimentation.

Scaling AI requires more than deploying another tool

An AI pilot can often be managed by a small team. Enterprise adoption cannot.

Once AI moves into multiple departments, the organisation needs decisions around data access, security, governance, integration, ownership, human oversight and measurement.

Technology teams need to understand how AI capabilities interact with existing applications and data environments. Business leaders need to determine where AI should influence decisions and where human judgement remains essential.

The organisation also needs a way to decide which AI initiatives deserve investment.

This becomes particularly important as AI spending increases. Gartner forecasts worldwide AI spending of approximately $2.7 trillion in 2026, representing 49.5% year-over-year growth. At the same time, McKinsey reports that one in five respondents say their organisations are limiting AI use because of operating costs.

The economics therefore matter alongside the technology.

A business may have access to increasingly capable models, agents and development tools, but every capability introduces questions around usage, infrastructure, integration, security and ongoing operating costs.

The organisations that benefit most will be the ones that can connect those costs to meaningful business outcomes.

The build-versus-buy decision is changing

AI is also changing how technology leaders think about software itself.

McKinsey's latest research found that 32% of respondents said their organisations had decided against purchasing at least one software product or feature because they believed they could build the functionality internally using agentic coding tools.

That is significant because it changes the conversation between technology and the business.

Historically, the decision might have been whether to purchase an existing application or commission a traditional development project. Increasingly, organisations have a third option: build a smaller, purpose-specific capability using AI-enabled development tools.

That does not mean every organisation should start building its own software.

The decision still needs to consider total cost of ownership, security, maintainability, integration, scalability, internal capability and the strategic importance of the capability being created.

But the economics of building are changing, and technology leaders need a framework for making those decisions.

AI architecture follows business architecture

One of the easiest mistakes is to begin an AI programme by choosing a model or platform.

The more important starting point is the work itself.

Where are people spending time? Which decisions are repetitive? Where are delays occurring? Which processes depend on information being manually gathered and interpreted? Where do employees repeatedly move information between systems? Where are customers experiencing unnecessary friction?

These questions reveal where AI may create value.

They also reveal where AI cannot solve the underlying issue.

If data is fragmented, the answer may involve data architecture before AI. If a process is poorly designed, automating it may simply make a poor process faster. If responsibilities are unclear, an AI agent will not solve the organisational problem.

This is why AI strategy increasingly needs to sit alongside operating-model design.

What should leaders be asking now?

For organisations moving from experimentation towards scale, several questions become increasingly important.

Which AI use cases are producing measurable business value?
Which pilots should become operational capabilities?
What changes to workflows are required for AI to create that value?
Where should AI make recommendations, and where should it be allowed to act?
What data does each use case depend on?
How will people supervise, challenge and improve AI-generated decisions?
What does each AI capability cost to operate at scale?
Who owns the outcome once an AI-enabled process becomes part of day-to-day operations?

And perhaps most importantly, what should the organisation stop doing because AI has changed the economics of the work?

These questions move the conversation from AI adoption to business design.'

The next stage of AI maturity

AI maturity in 2026 is increasingly about the organisation surrounding the technology.

The organisations that create lasting value will need the foundations to support AI, including reliable data, appropriate architecture, governance, security and clear ownership. They will also need leaders who are willing to redesign workflows rather than simply insert AI into existing processes.

That does not require every organisation to become an AI-first company.

It requires a clearer understanding of where AI changes the economics or effectiveness of work, what needs to change around it, and how that change can be sustained.

AI is becoming part of the operating environment of modern businesses. The next competitive advantage will come from knowing how to redesign the business around what that makes possible.

For technology leaders, that means the AI conversation is becoming broader than models, tools and pilots. It is becoming a conversation about how the business works.

Sources used