AIRI: A Practical Way to Assess AI Readiness

Organisations are rapidly adopting AI.

New tools and enterprise licences are being implemented, teams are testing copilots and agents, training programmes are expanding, and leaders are asking for AI use-case identification.

However, a fundamental question is often overlooked:

Are we truly prepared for these changes?

Readiness is not just about having access to AI.

It is not solely about whether employees have attended AI workshops.

Nor is it about having a few successful pilots.

AI readiness encompasses more than these factors.

It involves understanding where AI can add value, using it effectively, having the right data and infrastructure, maintaining strong governance, and ensuring AI initiatives link to tangible business outcomes.

This is why the evolution of the AI Readiness Index (AIRI) is particularly valuable.

AIRI: From AI Singapore origins to an open framework

Laurence Liew and his team at AI Singapore originally developed AIRI to help organisations understand their level of AI readiness and identify gaps.

Since then, the framework has developed independently.

Laurence Liew now maintains the current AIRI Framework v3.2 independently and releases it as an open standard under a Creative Commons CC BY 4.0 licence.

What interests me is not simply the assessment itself.

AIRI’s value lies in how it connects individual and organisational capabilities, governance, and business value.

These elements rarely develop at the same pace.

Employees may be AI-ready even if the organisation is not.

A key strength of the framework is its assessment of AI readiness at both individual and organisational levels.

The Personal AI Readiness Index, pAIRI, assesses individuals across five areas:

• Mindset

• Ethics & Responsibility

• Value Creation

• Data Literacy

• Tools & Technical Skills

It evaluates 15 dimensions, including AI understanding, learning agility, risk awareness, critical evaluation of AI outputs, use case identification, data literacy, specification thinking, and the ability to create or manage work using AI.

This distinction is important because:

Using AI does not mean you are AI-ready.

An individual may use ChatGPT, Claude, Gemini, or Copilot daily yet still struggle to assess output reliability, understand governance risks, determine appropriate AI applications, or demonstrate measurable value.

Frequency of use provides little insight into quality of use.

Organisational readiness is a separate consideration.

The Organisational AI Readiness Index (oAIRI) looks at readiness from an enterprise perspective.

Its five pillars are:

• Leadership & Culture

• Ethics & Governance

• Business Value

• Data Foundation

• Infrastructure & Standards

It examines factors including management support, workforce literacy, AI talent, a culture of experimentation, governance, risk controls, value realisation, data quality, infrastructure, and deployment standards.

This contrast highlights a critical point.

Consider an organisation where employees actively experiment with AI, build agents, and redesign workflows, yet the company lacks strong governance, adequate data access, and enterprise integration.

In this case, the workforce may outpace the organisation.

Conversely, consider the opposite scenario.

The company may invest in enterprise AI platforms, governance frameworks, and licences, while employees primarily use AI for writing or search assistance.

Here, the organisation may be technically prepared, but the workforce is not.

This underscores one of AIRI’s most valuable insights:

AI maturity is not a single metric; it is an alignment challenge.

AI readiness is also context-dependent.

Another strength of AIRI is its recognition of individual and organisational differences.

On AIRIhub, users can select their industry and, for the personal assessment, their functional role.

The underlying 15 dimensions and scoring remain consistent, but the language and examples are adapted to the work being assessed.

Industry options span areas such as healthcare and life sciences, technology, financial services, education, manufacturing, retail, logistics, construction and real estate, energy, professional services, government, telecommunications, hospitality, pharmaceuticals, legal services, insurance, agriculture and others.

The functional role options include leadership and management, operations, HR, finance, sales and marketing, IT, legal and compliance, engineering, clinical roles, educators, frontline teams, production, safety and quality, and creative roles.

This distinction is significant.

AI readiness should be defined differently for each group.

A marketer may need to understand AI-assisted research, content, customer data and brand risk.

A finance professional may care more about accuracy, controls and auditability.

A technical team may need deeper capability around infrastructure, integration and deployment.

A common framework ensures consistency.

Industry and role context make it more relevant.

This reflects a broader shift in approach.

Organisations are moving from generic AI training toward role-based and industry-specific AI capability development.

Rather than asking:

“Has everyone completed AI training?”

A more effective question is:

“Does each group understand how AI changes its work, risks, responsibilities and opportunities?”

AI literacy is only the foundation.

There has been a lot of emphasis on AI literacy in recent years.

That remains important.

However, literacy is increasingly viewed as the baseline rather than the goal.

A key advantage of pAIRI is its focus beyond basic AI understanding.

It covers areas such as specification thinking, including the ability to define outcomes, constraints, and checkpoints required for effective AI task completion.

It also assesses the ability to delegate work to AI while providing oversight, and to critically evaluate AI-generated outputs.

These capabilities are increasingly important as organisations transition from copilots to agents.

Prompting AI systems was valuable in the early stages of generative AI.

Now, understanding what to delegate, how to specify tasks, how to supervise, and when to withhold trust is even more critical.

Governance is integral to readiness, not an afterthought.

AIRI v3.2 is notable for its approach to governance.

Many maturity models treat governance as one factor among many.

AIRI introduces an Ethics & Governance gate, meaning an individual or organisation cannot achieve a higher overall maturity level than its governance level allows.

This is a valuable principle.

An organisation should not be able to compensate for weak governance simply by having excellent infrastructure, enthusiastic employees or sophisticated AI tools.

In summary:

Capability without governance does not constitute readiness.

As organisations adopt AI agents that can take action rather than only generate content, this distinction becomes increasingly important.

Higher maturity is not always the objective.

Another notable aspect of AIRI is that it does not assume all individuals or organisations should pursue the highest maturity level.

Its five levels range from AI Unaware, through AI Aware, AI Ready, AI Competent, to AI Catalyst.

But the framework suggests that AI Ready should be the target for most individuals and organisations.

This serves as a helpful reminder.

A professional-services firm may need a very different level of AI capability from an AI-native software company.

A marketing professional does not need the same depth as an ML engineer.

Organisations should not develop complex AI infrastructure solely because a maturity framework implies that higher levels are inherently better.

Readiness should align with organisational ambition.

It should not be the reverse.

Assessment is valuable, but action matters more.

Measuring readiness alone does not drive change.

AIRI’s greater value lies in linking readiness assessment to project selection.

Its AIRI Decision Engine considers areas such as leadership, governance, business value, data and infrastructure when evaluating potential AI projects.

This approach encourages a different sequence than many organisations currently follow:

Find a tool → run a pilot → search for a business case.

A more disciplined approach would be:

Understand readiness → identify real problems → assess value and feasibility → evaluate governance → prioritise → deploy → learn → reassess.

This may seem less exciting than launching another AI pilot.

However, it is likely the foundation for sustainable AI adoption.

Every AI strategy should begin with an honest baseline.

The enterprise AI conversation has been dominated by capability.

What can the latest model do?

Which agent platform should we deploy?

Which tasks can we automate?

Which vendor should we choose?

Those are important questions.

As AI capabilities advance, the primary constraints may shift to other areas.

• Leadership.

• People.

• Data.

• Governance.

• Business processes.

Organisational willingness to redesign work.

Organisations must also distinguish between impressive AI demonstrations and solutions that deliver sustainable business value.

The next phase of enterprise AI will not be defined by how many employees have access to AI.

It will depend on how effectively organisations align people, technology, data, governance, and business value.

An assessment cannot solve those problems.

But an honest assessment can expose where they are.

This provides a much stronger starting point.

Resources

You can explore the framework and assessments here:

• AIRI Foundation

• AIRI Framework downloads

• Personal AI Readiness Index, pAIRI

• AIRIhub pAIRI assessment

• AIRIhub oAIRI assessment