Why AI Investments Need Stronger Business Architecture

Organisations are investing millions in AI tools, platforms, infrastructure and consulting services, expecting improvements in productivity, operational efficiency and time to market.

However, these investments often don't match the internal effort required to translate AI capabilities into meaningful operational change.

Employees are encouraged to adopt AI, managers are expected to demonstrate productivity improvements, and technology teams are tasked with integrating AI into existing systems.

But an important question often remains unanswered:

Who is responsible for translating the organisation's AI strategy into how work actually gets done?

This is where Business Architecture needs greater attention.

A Gap That Existed Before AI

The disconnect between organisational strategy and everyday execution is not new.

Established frameworks such as TOGAF (The Open Group Architecture Framework) and the BIZBOK (Business Architecture Body of Knowledge) Guide have long recognised the importance of connecting business strategy with capabilities, value streams, processes, organisational structures and technology.

TOGAF provides a structured Architecture Development Method covering architecture vision, business architecture, information systems, technology, migration planning and implementation governance. BIZBOK provides guidance on business capabilities, value delivery, organisational alignment, and transformation planning.

Similarly, ArchiMate provides a modelling language for representing relationships across business, application, technology, strategy and implementation domains.

The architectural foundations already exist.

Yet having established frameworks does not necessarily mean organisations have successfully operationalised them.

In practice, Enterprise Architecture functions can become heavily focused on technology landscapes, application portfolios, infrastructure and solution governance. Business Architecture responsibilities may remain fragmented across strategy teams, transformation programmes, product management and operational leadership.

Consequently, organisations do not always consistently own the translation of strategy into clear operational direction.

The gap is not necessarily in the frameworks, but in how organisations implement and empower Business Architecture.

AI makes this existing weakness more consequential.

Unlike many traditional technology initiatives, AI can simultaneously influence how employees perform activities, how decisions are made, how products are developed and how business processes operate.

An organisation may have an AI strategy, approved platforms, governance policies, and technology roadmaps, yet lack a coordinated plan that explains how these investments should reshape processes, responsibilities, and daily work.

AI does not require organisations to reinvent Enterprise Architecture. It requires them to strengthen the connection between architecture, execution and measurable business outcomes.

When AI Adoption Becomes an Individual Responsibility

Without clear architectural direction, AI adoption increasingly becomes the responsibility of individual employees and managers.

Employees experiment with different tools, develop their own approaches and decide where AI fits into their activities. Managers introduce productivity expectations based on what their teams appear capable of achieving.

While experimentation is valuable, relying on it as the primary adoption mechanism creates inconsistencies.

Different departments may adopt tools with overlapping capabilities. Employees may use AI for activities where traditional automation would be more effective. AI-generated outputs may require additional review, correction and validation.

Some employees become highly productive, while others struggle to integrate AI into existing processes.

More importantly, individual productivity improvements do not automatically translate into organisational productivity.

An employee completing a task faster does not necessarily mean the organisation delivers a product or service faster.

If AI accelerates content creation, development or analysis but increases the effort required for validation, integration or downstream processing, the expected benefits can diminish.

The organisation may produce more output without generating proportionate business value.

Productivity Gains Must Be Measured Across the Business

One of the primary expectations of AI adoption is a shorter time to market.

AI can accelerate requirements analysis, design, development, documentation, testing and many other activities. However, these activities are interconnected.

Consider a product development process where AI significantly reduces development effort. If the generated solution introduces integration issues, additional testing requirements or long-term maintenance complexity, the overall delivery time may remain unchanged or even increase.

Similar situations can emerge in non-technical functions.

Faster document preparation may increase the review burden. More automated analysis may create additional verification work. Greater output from one department may overwhelm another.

The problem is not necessarily the AI tool.

The problem is the lack of a coordinated approach to redesigning the entire business process around AI capabilities.

Organisations should evaluate productivity across end-to-end value streams rather than individual activities alone.

This requires understanding where AI creates value, where human judgement remains essential, and where introducing AI simply shifts effort or risk elsewhere in the organisation.

Without this coordination, AI adoption can unintentionally increase operational costs and extend delivery timelines—the opposite of its intended purpose.

Business Architecture Must Connect Strategy to Daily Work

Business Architecture should take on a more prominent role in organisational AI adoption.

This does not require creating a parallel AI architecture framework. Instead, organisations should extend established Business Architecture practices to address AI-driven changes in work, decision-making, investment, governance and organisational capabilities.

The Business Architecture function should help translate AI strategy into practical operating decisions:

• Which business capabilities genuinely benefit from AI?

• Which activities should be automated, augmented or remain human-led?

• How should processes, roles and responsibilities change?

• What principles should guide employees in choosing and using AI?

• Where must human judgement, accountability and controls remain?

• What skills and capabilities do employees need?

• How should productivity improvements influence workload and delivery expectations?

• How should AI investments be connected to measurable business outcomes?

This is not about introducing another management layer or creating additional approval processes.

It is about establishing a coherent operating framework that departments and employees can follow.

Managers should continue to adapt and implement changes within their teams. Technology architects should remain responsible for technical design. HR, finance, security, governance and operational functions should retain their respective responsibilities.

Business Architecture should connect these decisions so that individual initiatives contribute to the same organisational objectives.

AI Cost Optimisation Is More Than Technology Cost

Another important consideration is the growing operational cost of AI.

Beyond initial investments, organisations incur recurring expenses through licences, model consumption, infrastructure, integrations, monitoring, validation and ongoing maintenance.

Without a coordinated architectural view, departments may independently adopt similar tools or integrate AI into processes without evaluating whether those investments are justified.

Not every activity requires a large language model. Not every business process needs an AI agent. In some cases, conventional automation or process improvements may deliver better value.

AI cost optimisation therefore requires more than selecting cheaper models or negotiating licensing agreements.

It requires understanding business needs, identifying reusable capabilities, avoiding unnecessary duplication and evaluating the total cost of achieving an outcome.

Business Architecture can provide this business-level perspective by working with technology, finance, and procurement.

The objective should not be to maximise AI utilisation, but to maximise the business value generated from AI investment.

High AI consumption is not necessarily evidence of successful transformation.

Integrating AI Into Products Introduces Another Responsibility

Employee adoption is only one part of organisational AI transformation.

Organisations are also integrating AI into products, customer experiences and core business processes.

Such integration can create greater business value than standalone tools, but it introduces additional responsibilities and risks.

AI capabilities can affect data privacy, security, regulatory compliance, reliability, product behaviour, customer trust and decision accountability.

A technically successful integration may still be a poor business decision if it increases operational costs, creates unacceptable risks or compromises customer experience.

This is where Business Architecture must work closely with product, engineering, security, legal, compliance and operational teams.

It should help establish where AI integration makes business sense, how processes will change, and which operational boundaries must be maintained. Relevant technical and risk specialists remain accountable for designing and assessing controls within their domains.

The objective is to balance innovation, business value, cost, and risk without letting departments optimise their individual priorities at the organisation's expense.

The goal is not to integrate AI everywhere it is technically possible, but where it creates justified business value without compromising essential business requirements.

From AI Adoption Metrics to Business Outcomes

Many organisations track AI adoption through licences activated, employees using tools, AI use cases deployed or processes incorporating AI.

These metrics provide useful information about adoption, but they do not necessarily establish business value.

An organisation may report increasing AI usage while still experiencing higher operational costs, additional validation effort or limited improvements in end-to-end delivery.

A more meaningful assessment should ask whether AI has:

• Reduced overall delivery time rather than individual task duration.

• Improved the quality and reliability of business outcomes.

• Lowered the total cost of delivering a product or service.

• Increased operational capacity without disproportionate workload or risk.

• Improved customer experience or created measurable business value.

Business Architecture can help establish traceability between AI investment, the business capabilities being improved, changes in operational processes and the benefits actually realised.

Without this connection, organisations risk confusing increased AI activity with successful AI transformation.

Designing the Organisation Around AI

AI adoption is frequently presented as a technology transformation supported by employee training and change management.

But its implications are broader.

AI influences how work is distributed, how employees collaborate, how decisions are made, how products are developed and how business capabilities are delivered.

These changes require deliberate organisational design.

This does not necessarily mean every organisation needs a new or larger Business Architecture department. What matters is having a clearly empowered capability that translates strategy into coordinated operational decisions across existing business and technology functions.

In my previous article, I discussed employee resistance to AI adoption and the uncertainty surrounding productivity expectations, job security and changing responsibilities.

Some of that uncertainty exists because employees are being encouraged to adopt AI without clear guidance on how their work, responsibilities, and expectations will evolve.

Stronger Business Architecture can help organisations address these questions systematically rather than leaving individual managers and employees to interpret the AI strategy themselves.

Organisations should invest not only in making AI available to employees, but also in designing how AI should improve the work they perform.

Ultimately, successful AI transformation is not about how many employees use AI or how many AI capabilities are deployed.

It is about whether those investments produce measurable improvements in productivity, cost efficiency, time to market and business outcomes.

AI investments create capabilities. Business Architecture helps turn those capabilities into organisational value.

The question organisations should be asking is no longer just “Where can we introduce AI?”

It is also:

“Who is responsible for ensuring that AI changes the way our organisation works for the better?”