Putting things in context

The conversation around artificial intelligence swings between breathless hype about machines replacing entire professions and deep scepticism about whether any of it can be trusted. The truth sits somewhere in the middle.

AI isn't magic, it's a powerful tool that responds to the context it's given. Understanding that changes how we use it safely and professionally. The answer you get will only be as good as the prompt or instruction you write. When people talk about prompt engineering, they mean giving the AI clear instructions to enable it to produce better results.

Think of it less like programming a computer and more like briefing a colleague. If you were to say: "Write something about cremations", you'd get a very different result than if you said: "Please draft a short explanation of the cremation process for a family that has never arranged a funeral before. Use plain language and a reassuring tone."

The second instruction provides context – the audience, the purpose and the tone – and context is everything. Across AI, people are moving from the term 'prompt engineering' towards 'context engineering', which is more accurate.

Think of it less like programming a computer and more like briefing a colleague.

You might use AI to help draft obituary notices, service scripts, family information packs and website articles explaining funeral options. The more guidance you can give it, the better. For instance: "Using the tone and language from our previous client communications, write a gentle explanation of the difference between direct cremation and a simple attended funeral for families who may be unfamiliar with both options. Base your explanation on the information in our service guide (attached), ensuring it reflects how we describe both services in practice. Use respectful, clear language, avoid promotional phrasing, and ensure the explanation aligns with our approach to supporting families without influencing their decision."

If you ask a general-purpose AI about funeral arrangements, it will draw on generic knowledge from the internet, which will be broadly correct, but won't reflect your firm's procedures, your local practices, or the professional standards you follow.

If the AI can access your actual company documents – your templates, policies and guides – the response becomes far more reliable because it's grounded in the truth of your organisation. When AI can draw from your real documents, it becomes more useful and far less prone to inventing things.

AI works best given clear, detailed instructions, the right documents, defined boundaries and, finally but crucially, human oversight. AI is excellent at augmentation, but no machine can understand the emotional nuance of a bereaved family the way an experienced funeral director can. No algorithm can substitute for the discretion required when navigating delicate family dynamics or cultural expectations. And that is precisely why governance matters regardless of how good your prompt is.