Some not-for-profits try ChatGPT once, get a generic answer, and quietly decide: “This isn’t for us.”
Often, the issue isn’t the tool. It’s that ChatGPT hasn’t been told who you are, who you serve, or how you want it to respond.
There’s a small setting called Personalisation (also known as Custom Instructions). It takes about five minutes to set up, and it can significantly improve how useful ChatGPT feels for NFP work, especially where tone, clarity, and trust matter.
OpenAI itself describes Personalisation as a way to share preferences and context once, so you don’t need to repeat them in every conversation. In practice, that means ChatGPT can start each task already aware of your role, audience, and tone, rather than guessing or defaulting to something generic.
I use this myself for drafting emails and LinkedIn articles. Instead of re-explaining my context each time, I’ve set it up so ChatGPT already knows the style I’m aiming for and the audience I’m writing to. The first draft comes back closer to what I actually need, which means less rewriting and less cognitive load.
What goes into Personalisation?
Think of it as two short prompts you set once:
- What ChatGPT should know about you
- Your role and sector (for example, working in not-for-profit or community services)
- Who you usually write for (staff, boards, funders, community)
- Key constraints, such as limited time and the need for plain English
- How you want ChatGPT to respond
- A clear tone (warm, grounded, non-technical)
- Preferred structure (short paragraphs, dot points when helpful)
- Guardrails: flag uncertainty, don’t invent facts or quotes, ask when more context is needed
This matters for more than efficiency.
From an ethics and risk perspective, Personalisation supports:
- Consistency in organisational voice
- Transparency about uncertainty
- A clear human-in-the-loop approach
If your team has “tried AI” and it felt off-brand or unhelpful, this is the first fix I’d suggest before investing in new tools.
Add comment
Comments