An AI strategy is the organisation's approach to using AI to support its objectives. It should explain where AI can create value, how investments will be prioritised, what capabilities are needed and how progress will be governed and measured.

Start with business strategy

The 2026 AICD and Human Technology Institute Director's Guide to AI Governance recommends that boards discuss the strategic opportunities AI presents and, where appropriate, develop an AI strategy with management. It warns against adopting tools that are disconnected from organisational strategy or expected return.

This means the first question is not which AI product to buy. Start with the business outcomes that matter, such as service quality, workforce capacity, customer access, operational resilience, revenue growth or better use of organisational knowledge.

AI should act as an enabler of organisational strategy. A separate list of AI projects is not enough.

Build a baseline before setting the roadmap

The AICD and HTI recommend a stocktake or mapping exercise before an organisation develops and implements its AI strategy. The baseline should show:

  • where AI is already used, including embedded supplier features;
  • current skills, data, technology and supplier dependencies;
  • active pilots and planned investments;
  • known costs, benefits and performance evidence;
  • material governance gaps and information that remains unknown.

Structured discovery gives leaders an honest starting point. It can reveal useful work worth scaling, duplication that should be consolidated, and high-impact uses that need stronger review.

Define a small set of strategic objectives

Each objective should describe an outcome, not a technology. For example:

  • improve the quality and consistency of customer service;
  • reduce avoidable administrative work while maintaining service standards;
  • strengthen how organisational knowledge is found and reused;
  • improve planning and operational decision support;
  • develop workforce capability for responsible AI adoption.

For each objective, define the intended beneficiaries, accountable executive, time horizon and how success will be measured.

Set principles and risk appetite

Strategy needs boundaries. Agree the organisation's principles for responsible AI use and the types of impact that require greater scrutiny or executive approval. Risk appetite may differ across uses. An internal drafting aid and an AI-supported decision affecting a person's employment should not follow the same path.

The principles should be reflected in investment decisions, assessment questions, controls and approval pathways. Otherwise they remain statements with no operational effect.

Turn objectives into a prioritised roadmap

A roadmap translates strategy into sequenced work. Group the work into three categories:

  1. Foundation initiatives. Discovery, policy, data quality, skills, governance roles and the AI register.
  2. Value initiatives. Defined AI opportunities or use cases linked to strategic objectives and measurable outcomes.
  3. Enabling initiatives. Technology, data, supplier, workforce and change capabilities required to deliver the portfolio.

Prioritise initiatives using consistent criteria such as strategic alignment, expected value, feasibility, readiness, impact on people, data dependency, cost and uncertainty. Keep proposed ideas visible without treating every idea as an approved project.

Measure progress at three levels

  • Strategy progress: objectives on track, roadmap milestones completed and capability improvements delivered.
  • Portfolio progress: initiatives moving from idea to pilot to scaled use, with governance readiness and investment status.
  • Use-case value: specific measures of cost, time, quality, access, revenue, safety or service outcomes for each use where reliable evidence exists.

The AICD and HTI recommend agreed measures and indicators of AI progress and ultimate success, including return on investment. They also recommend measuring returns at the level of specific use cases.

CSIRO similarly describes AI project evaluation as a blend of financial, ethical and strategic thinking. This is useful because the cheapest or fastest proposal may not be the best strategic choice.

Avoid false precision

Not every benefit can be reduced to a dollar figure. Use financial measures where the evidence supports them, and keep service, quality, workforce and stakeholder outcomes visible as separate measures.

Connect strategy to governance

Each roadmap initiative and registered AI use should link to a strategic objective. Governance information then becomes part of portfolio management:

  • Is the use aligned with strategy and risk appetite?
  • Does it have an accountable owner and a credible business case?
  • What must be resolved before pilot or deployment?
  • Is the expected value being measured?
  • Did a material change alter the risk or business case?
  • Should the organisation scale, adjust, pause or retire the use?

A concise AI strategy structure

  1. Organisational context and the role AI will play.
  2. Strategic objectives and intended outcomes.
  3. Current-state discovery and capability baseline.
  4. Principles and risk appetite.
  5. Priority opportunity areas and selection criteria.
  6. Roadmap of foundation, value and enabling initiatives.
  7. Accountability, investment and decision pathways.
  8. Measures, reporting cadence and review process.

The strategy should be reviewed as business priorities, AI capability, supplier offerings and evidence change. Progress belongs in the same management and board reporting cycle as governance status.

Sources and further reading

This article provides general strategy and governance guidance. Organisations should adapt the approach to their objectives, sector, operating model, capability and risk profile.