thinkhard.SI

Prompt repository Managers & leads

Introduce an AI tool to your team with clear rules

Your team is starting to use an AI tool, and you want clear rules on what it is for, what data never goes in and how output gets checked.

Registry no.
MGT-051
Collection
Managers & leads
Laid out for
Any chat model
Score
10 out of 10
Blanks to fill
7
Added

Managers & leadsAny chat

The prompt

Score 10 of 10, Excellent

Use it when Your team is starting to use an AI tool, and you want clear rules on what it is for, what data never goes in and how output gets checked.

  • {{tool}}
  • {{team}}
  • {{policy}}
  • {{tasks}}
  • {{approval}}
  • {{concerns}}
  • {{training_time}}
Full prompt and why it works
Prompt
Help me introduce {{tool}} to {{team}} with clear rules, so that people use it for the work it suits, never put in data our policy forbids, and check its output before anything leaves the team.

Our company's policy on AI tools and data, as written:
"""
{{policy}}
"""

What the team does, and the tasks I think the tool could help with:
"""
{{tasks}}
"""

What the tool is approved for, and how people get access: {{approval}}
Concerns I have heard from the team: {{concerns}}
Time we can give to training: {{training_time}}

Produce, with a heading for each part:
1. The team guide, under 500 words, to share with everyone, with these short sections:
   - What to use it for: the tasks from my list that fit, each with one line on how.
   - What not to use it for, and why.
   - Data that must never go in, taken from the policy.
   - Review rules: what must be checked before output is used, and by whom, for example every fact, number and name in anything sent outside the team.
   - Telling people when AI helped, as far as the policy requires it.
   - Sharing good prompts: one shared place, with the prompt, what it is for and an example of the output.
   - Who to ask when unsure.
2. A rollout plan for the first six weeks that fits the training time above, as a table with the columns Week, What happens, Who, Done when.
3. What to measure: up to four measures, such as time saved on a named task by self-report or errors caught in review, with how and when each is collected.
4. Answers to the team's concerns, short and honest.

The guide should read like a practical team agreement, not a policy document: short lines, plain words, real tasks.

Take every data rule from the policy, never from general practice, because the policy is what people will be held to. Do not claim the tool is accurate, secure or compliant beyond what the approval says. Do not promise that AI will not affect anyone's job unless the company has said so; if that concern has come up and there is no company statement, the answer is that I will find out and come back to them.

If the policy does not cover something the guide needs, such as whether client documents may go in, write [POLICY GAP: question for IT or legal] and treat it as not allowed until answered. If there is no policy at all, stop and list the questions to settle before rollout.

Variables

{{tool}}
The AI tool, for example "our company's approved chat assistant".
{{team}}
Your team, for example "the eleven people in client services".
{{policy}}
Your company's AI or data policy, pasted as written. Write "none" if there is no policy.
{{tasks}}
What the team does, and the tasks you think the tool could help with.
{{approval}}
What the tool is approved for and how people get access, for example "approved by IT for internal documents; licenses through the help desk".
{{concerns}}
Concerns the team has raised, or "none yet".
{{training_time}}
How much time you can give to training, for example "one hour a week for four weeks".

Why it works

  • What to avoid. Every data rule comes from your policy, not general practice, and no claim about the tool's accuracy or security goes beyond its approval.
  • If unsure. A gap in the policy becomes a [POLICY GAP] question and counts as not allowed until it is answered, so the guide never permits something by default.
  • Format. A one-page guide, a six-week table and a short list of measures give the team rules to follow and you a way to see whether the tool helps.

How to use it

  1. Press Fill in blanks and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.
Use it from code

Get this prompt as JSON from the free REST API, with your own text in the blanks, or let an AI assistant fetch it through the MCP server. No key.

curl -s https://thinkhard.si/api/v1/prompts/team-ai-rollout-plan

With the MCP server added to Claude, ChatGPT, Cursor or VS Code, call get_prompt with {"prompt": "team-ai-rollout-plan"}, or just ask for it by name.