New year, new budget cycle, new pressure to have an “AI strategy.” If you’re in a planning conversation right now, here’s a framework for thinking through 2025.

Start With Business Problems, Not AI

This remains the most common mistake: starting with “we need to do more AI” rather than “we need to solve these business problems.”

Before any AI planning, answer:

1. What are our top 3-5 business challenges for 2025?
2. Where are we losing money, time, or customers?
3. What competitive pressures require response?
4. Where could technology (not specifically AI) help?

Only after answering these should AI enter the conversation. Some of your challenges may benefit from AI. Some may need simpler solutions. Some may not be technology problems at all.

Audit What You Already Have

Most organisations have more AI deployed than they realise:

• Microsoft 365 Copilot and AI features
• CRM tools with embedded AI
• Marketing automation with predictive capabilities
• Finance tools with AI-assisted forecasting
• Security tools with ML-based detection

Before planning new initiatives, understand what you have:

• What AI is deployed across the organisation?
• What’s being used effectively? What’s shelfware?
• What’s the total current spend on AI tools?
• What governance exists for these tools?

This audit often reveals opportunities to extract more value from existing investments before adding new ones.

Define Your AI Maturity Honestly

Not every organisation should pursue advanced AI. Your strategy should match your maturity:

Foundational (most organisations):

• Focus: Productivity tools, basic automation, data foundations
• Investments: Copilot-style tools, data quality, governance basics
• Risk profile: Lower, proven applications

Developing:

• Focus: Custom applications for specific business processes
• Investments: ML platforms, data science capability, domain-specific AI
• Risk profile: Medium, requires proven data and skills

Advanced:

• Focus: AI as competitive differentiator, embedded in products/services
• Investments: Custom model development, AI-native products, advanced capabilities
• Risk profile: Higher, requires significant capability

Be honest about where you are. Pursuing advanced strategies with foundational maturity is a recipe for expensive failure.

The 2025 Priority Stack

Based on patterns across Australian enterprises, here’s a suggested priority stack:

Priority 1: Extract Value From Existing Tools

You’ve probably already invested in AI-enabled tools. Are they being used effectively?

Actions:

• Assess adoption rates for tools like Copilot
• Identify training gaps and address them
• Optimise configurations based on actual usage patterns
• Measure ROI and adjust licensing accordingly

This is often the highest-ROI work: getting value from investments already made.

Priority 2: Strengthen Data Foundations

AI is only as good as the data it uses. For most organisations, data work should be a 2025 priority.

Actions:

• Assess data quality for key business domains
• Establish data governance for AI-relevant datasets
• Improve data integration between systems
• Address data access and consent issues

Not glamorous, but essential. AI strategies built on weak data fail.

Priority 3: Build Governance and Risk Management

As AI becomes more prevalent, governance becomes non-negotiable.

Actions:

• Establish AI governance framework and accountability
• Define risk tiers and review processes
• Address EU AI Act implications (if applicable)
• Create policies for AI use, particularly generative AI

Getting this right prevents painful cleanup later.

Priority 4: Pursue High-Value Use Cases

With foundations in place, pursue specific AI applications:

Actions:

• Identify 2-3 high-value use cases with clear business cases
• Pilot with appropriate rigour
• Scale what works, learn from what doesn’t
• Build internal capability alongside external delivery

The temptation is to pursue many use cases. Resist it. Depth beats breadth.

Priority 5: Develop AI Capability

Building internal capability reduces long-term dependency on external providers.

Actions:

• Upskill existing staff on AI fundamentals
• Recruit selectively for critical skills gaps
• Partner with universities or training providers
• Create communities of practice to share learning

This is a multi-year effort, but 2025 should advance it.

Budget Allocation Guidance

How should budget split across these priorities? A reasonable starting point:

Priority	Budget Share
Existing tool optimisation	15-20%
Data foundations	25-30%
Governance and risk	10-15%
New use cases	30-35%
Capability development	10-15%

Adjust based on your maturity. Less mature organisations should invest more in foundations; more mature organisations can allocate more to use cases.

What Not to Do in 2025

Some patterns to avoid:

Chasing every shiny object: New AI capabilities will emerge monthly. You can’t pursue all of them. Stay focused on your prioritised use cases.

Underinvesting in change management: The pattern of technology success and adoption failure continues. Budget for change management upfront.

Ignoring governance: Regulatory scrutiny is increasing. Organisations without governance will face problems.

Expecting quick wins: Meaningful AI takes time to deliver value. Set expectations accordingly with stakeholders.

Going it alone: Unless you have significant internal capability, you’ll need partners for complex AI work. Budget for external support.

The Planning Process

Practical steps to develop your 2025 AI strategy:

January-February: Audit current state (tools, data, governance, capability)

February-March: Identify priority use cases, build business cases

March-April: Develop roadmap, secure budget approval

Q2: Begin implementation of highest-priority initiatives

Q3-Q4: Scale what works, adjust what doesn’t, plan for 2026

Don’t try to plan the whole year in detail. Technology evolves too fast. Plan Q1-Q2 specifically, Q3-Q4 directionally.

Measuring Progress

Define metrics before starting:

Adoption metrics: Are tools and capabilities being used?

Value metrics: Are business outcomes improving?

Capability metrics: Are skills and practices developing?

Risk metrics: Are governance and compliance on track?

Review quarterly and adjust strategy based on what you learn.

Final Thought

AI strategy doesn’t need to be complicated. It needs to be:

• Grounded in business problems
• Realistic about maturity
• Focused on a few priorities
• Supported by appropriate investment
• Measured and adjusted over time

The organisations that succeed with AI in 2025 won’t be the ones with the most ambitious strategies. They’ll be the ones that execute well on focused, realistic plans.

Start simple. Build capability. Expand gradually. That’s the path that works.