AI Prompt Engineering

Start by clearly stating who the AI is supposed to be (its role) and who it's helping (the audience).

The role could be a specific job function (like a marketer or an HR manager), a job level (like an entry-level professional or a C-Suite executive), or other persona you want the AI to embody.

The audience is who the output is for. Who will consume or benefit from what you’re creating?

This information helps shape how the tool solves your problem or responds to your request.

Next, tell AI what task you want it to do or what action you want it to take. Be as specific as possible about the steps you want it to follow or the outcome you’re trying to achieve. (Vague requests like “Help me…” won’t produce great results.)

Then, list any requirements or preferences you have in mind for the AI’s answer. This could include:

* The tone or style you want the writing to be in

* How long you want the response or deliverable to be

* The format it should be in (such as an email, table, or bullet points)

Again, be as specific as possible. Unlike humans, AI doesn’t get overwhelmed or offended by too many details.

Like a teammate, AI needs context to produce relevant, useful responses. Give the tool as much relevant background information as you can think of.

Think about the “who, what, where, when, why, and how” of your task, such as:

• Who: Personas or details about the audience

• What: What you already know about the task, such as research or work that’s already been completed

• Where and when: Where and when the background information came from and/or where and when the output will be used

• Why: Why this task is important and the value you’re trying to provide

• How: How you want the AI to proceed, steps you want it to follow, and/or how the output will be used

Don’t settle for the first response

Treat your AI interaction as a conversation, and provide more details or instructions to refine the output. For example, if you asked for help creating a project plan but the timeline it provided was too aggressive, give that feedback and tell the tool to extend the project plan over a longer period of time.

Put AI to work

Keep the momentum going, and run other Plays to make AI part of your team:

1. AI Innovation Day: Build your team’s confidence with AI and explore more ways to use it to improve workflows.

2. AI Training Workshop: Run a workshop to help anyone learn the basics of how to use AI at work.

3. Fix It Friday: Schedule some time on a Friday to improve one workflow with AI.

4. AI Teammate: Build your first AI agent.