Context engineering

Definition

The design and management of the information, instructions, memory, tools and resources made available to an AI system.

Put simply

Prompting tells AI what you want. Context engineering determines what resources it has available to do the job well.

What boards and leaders should understand

As AI moves from simple conversations to longer-running work, the quality of the surrounding context becomes increasingly important.

The best model can still perform poorly if it receives outdated policies, irrelevant information, contradictory instructions or insufficient organisational knowledge.

AI capability can therefore depend not only on the model an organisation buys, but on how well the organisation can assemble and maintain its own knowledge around it.

Questions to consider

  • What proprietary organisational knowledge could make our AI capability better than a competitor using the same underlying models?