Building an AI Strategy

Most organisations are adopting AI tactically — deploying tools in response to immediate pressures — rather than strategically. A deliberate strategic approach produces better outcomes and lower risk.

Starting with the strategic question

Before technology selection, organisations should be clear on: Where does AI create sustainable competitive advantage for us specifically? What are our constraints (regulatory, data, talent, risk tolerance)? What AI capabilities are table stakes for our industry versus genuine differentiators?

The AI capability continuum

AI strategy typically evolves across three stages:

Adopter — using AI tools built by others (Microsoft Copilot, ChatGPT Enterprise, Salesforce Einstein). Low technical barrier, fast to value, limited differentiation.

Customiser — building on top of foundation model APIs with proprietary data, fine-tuning, and RAG. Requires engineering capability, produces domain-specific AI advantage.

Builder — training or fine-tuning models at scale on proprietary data. Requires significant ML capability, data, and compute. Only justified where the differentiation from custom models is genuinely significant.

Data strategy as AI strategy

The organisations that will derive the most value from AI are those with proprietary data advantages — unique data that competitors cannot easily replicate. Investing in data quality, data infrastructure, and data network effects (where more users generate more valuable data) is AI strategy at its most fundamental level.

AI strategy is not an IT strategy. The organisations getting the most from AI are those where the CEO and leadership team are driving it — not delegating it to the technology function.