The Data Front

How to brief your AI so it works with you, not for you

Once you’ve tried some basic prompting, the next step is learning to guide your AI more deliberately. This isn’t about writing long, complicated prompts – it’s about treating your request like a brief: clear, structured, and purposeful. Because the better you frame the task, the better the AI can help.

Using AI with intent matters; every unnecessary prompt or sprawling request carries an environmental cost, so the more purposeful we are in how we use these tools, the more value we gain without wasting resources.

---

Set the role

The first layer of a good prompt is perspective. AI works best when it knows who it’s meant to be in the moment.

You might start with:
“You are an internal communications lead”

But you’ll get stronger results if you add detail:
“You are an internal communications lead with experience running change programmes. You write in a clear, calm tone with a focus on empathy.”

That way the AI has a role, a level of expertise, and a tonal guide.

Build the persona

The next step is to expand that role into something fuller. This gives your AI a clear voice and direction, almost like briefing a colleague you trust.

You might include:

• Professional experience
• Knowledge of a platform, tool, or audience
• The style or tone you want
• The kind of impact the message should have

Here are two worked examples:

Example 1
You are a senior data enablement manager at a global retailer. You’ve launched internal learning programmes before, and you know how to explain change in a way that feels empowering. You’re drafting a short post to share a new data resource with internal teams. The message should feel practical, clear and optimistic.

Example 2
You are a visualisation lead who understands how to make design approachable. You’re writing for a mixed audience of analysts and business users. Your job is to introduce a new dashboarding resource in a way that feels inclusive, simple, and immediately useful.

Layer in context

Now add the background. What does your reader already know? Where will this be shared? Are there sensitivities to be aware of?

For example:

• The audience is a community Slack group and the company newsletter
• Readers have seen training before but often find it generic
• The aim is to encourage adoption without feeling pushy

Bring it all together

Here’s what a finished prompt might look like:

You are a senior internal enabler with experience launching learning programmes for Tableau users. You’re announcing a new self-paced e-learning pathway designed to help teams improve dashboard design and data storytelling. The audience is our internal data community – a mix of analysts and business users. The tone should feel inclusive, clear and optimistic. The message will go out in our community Slack and internal newsletter. Draft a short announcement that highlights what’s available, why it matters, and encourages people to check it out.

You could then follow up with:

• Make this more conversational
• Add a line about why this is different from past training
• Cut it by 30% but keep the impact

Two extra tips:

1. Ask the AI to help you build the prompt. For example:
   “Help me write a prompt for announcing a new data programme. I’ll fill in the details, you give me the structure.”
2. Add a line to your prompt that says:
   “Ask me any follow up questions you need to get this right.”

It makes the exchange collaborative, not one-sided.

---

Next in the series, we’ll zoom out to look at the Waves of AI – from traditional models, to generative tools, and the agentic future that’s just around the corner.