Talking points, tough questions and a practice run before the real conversation, so managers can explain a pay outcome clearly and honestly.
A pay outcome explained badly does more damage than the outcome itself. Managers blame HR, over-promise, or freeze when someone pushes back. Reward teams cannot sit in every conversation. This tool puts good preparation in every manager's hands.
It has two halves. The prep builder is ordinary code. The manager picks the outcome, the main reasons and how the employee is likely to react, and gets talking points, likely tough questions and what not to say. The same choices give the same output every time. The practice partner is a ready-to-paste prompt that makes the manager's own approved AI assistant play the employee, then give feedback.
The benefit is fewer escalations landing back on the reward team, and managers who sound like they understand the decision.
This is the real prep builder, running in your browser. Pick an outcome, up to three reasons and a likely reaction, then press Build my prep.
Nothing you choose is saved or sent anywhere.
This is what the practice prompt does in an AI assistant. The employee is frustrated. The manager lines are scripted. Press Play to watch it, or Show all to read it.
The practice itself happens in your own approved AI assistant, not on this site.
Five steps from a pay outcome to a rehearsed conversation. Press Play, or step through at your own pace.
Pick the situation. Three choices, no typing. No names and no pay figures, so there is nothing sensitive to enter.
Talking points you can say aloud. Four parts: say the outcome first, explain the decision, give the reasons one at a time, then say what happens next.
The questions you are dreading. The tool lists the questions this employee is likely to ask, with a suggested answer and the trap to avoid for each.
What not to say. Common lines that backfire, each with a better alternative and the reason.
Rehearse it. Paste the practice prompt into your approved AI assistant. It plays the employee, then steps out and gives feedback quoting your own words.
Everything in this walkthrough is an invented example, for illustration only.
You can build your own version with any capable AI assistant your employer allows, for example Microsoft Copilot, Claude or ChatGPT. You describe what you want, it writes the code, and you check the result.
The tool you build contains no AI. The wording is yours, so the most important step is the second one: giving the tool your organisation's own words, then having your reward lead review every line.
Say what the tool is, what goes in, what comes out, and who will use it.
The AI does not know your pay review process. Let it interview you, then write the content from your answers.
Have your reward lead review every line before any manager sees it. The AI can write the wording. It cannot know whether the wording matches your policy.
The fourth tab writes a prompt for the manager's own AI assistant. Be exact about what that prompt must make the assistant do.
Test the hardest situation first. Then read every line aloud.
Then do the real test yourself. Read every line aloud, and remove anything a manager should not say.
The prep builder touches nothing sensitive. It takes choices only, with no names and no figures. The practice run is where the risk is, because managers are tempted to type the real person's details to make the practice feel real.
Practise with the situation, not the person. Leave out names, real pay figures, and health or personal circumstances. Use only an AI assistant your employer has approved.
Privacy rules for employee records differ between private employers and government agencies, so check with your privacy officer. Start with the OAIC's guidance on the Australian Privacy Principles and on employment and privacy. General information, not legal advice.
It is a single file. Save it anywhere and double-click to open it. Before you share it with managers, edit the wording to match your organisation's policy and have your reward lead review it. The captured practice run is here: practice-transcript.md.
What next: Back to all use cases