Sharing what works across a team

What it is

In many teams, a few people find good ways to use AI and everyone else starts from scratch. Sharing what works means turning one person's good prompt or method into something the whole team can use.

The basic tools are simple. A shared library of prompts and saved instructions (sometimes called skills: a reusable set of instructions for one job). A short regular slot where people show what they tried. A few AI champions, who are colleagues that help others and keep the library tidy. Clear names, version numbers and credit for whoever wrote each item.

Think of a family recipe book. When someone makes a dish everyone loves, they write it down, put their name on it and note any changes over the years. Nobody has to rediscover how much salt to use.

Why it matters

• One good prompt, reused by ten people, saves far more time than one person using it alone.
• New team members get up to speed faster with tested examples.
• Showing real work builds confidence in a way that training slides cannot.
• Credit encourages people to share instead of keeping their tricks private.
• A shared library makes it easier to spot and fix a prompt that gives poor results.

How to do it

1. Create one shared place for prompts, such as a folder or a page in a tool your organisation already approves.
2. Agree a simple format for each entry: name, what it does, the prompt, an example, author, version and date of last change.
3. Name entries by task, such as "Meeting notes to action list", so people can find them.
4. Set up a 15-minute show-and-tell each week or fortnight where one or two people show something that worked, or did not.
5. Ask two or three volunteers to act as AI champions: they answer questions, review new entries and remove duplicates.
6. Set clear rules on what may be shared: prompts and made-up examples yes; customer, personal or confidential data never.
7. Measure adoption simply: how many entries, how often they are used, and what people say helped.

Example

For example, a sales support team had six people each writing their own prompts for quote follow-up emails. Results varied, and one person's version was clearly better.

The team added it to their shared library:

Name: Quote follow-up email
Version: 1.2
Author: [name]
What it does: drafts a polite follow-up to a quote sent [number] days ago.
Prompt:
Write a short, friendly follow-up email about the quote we sent on [date] for [product or service].
Ask if they have questions, offer a short call, and keep it under 120 words.
Do not mention discounts or change any terms.
Example input: [made-up customer and product]
Last change: added the "do not mention discounts" line after a review.

At the next show-and-tell, the author spent five minutes showing it. Within a month, most of the team used it, and each person still read and adjusted every email before sending.

Common mistakes

• Sharing real data by accident. Examples copied from real work can contain names or figures. Replace them with made-up details before sharing.
• A library nobody can find anything in. Use clear task names and remove duplicates regularly.
• No versions. When a prompt changes without a note, people cannot tell why results changed. Add a version number and a short change note.
• Only sharing successes. Hearing what did not work saves others time. Welcome both.
• Making it a burden. Long forms stop people contributing. Keep the format short.

Try it this week

1. Choose one prompt you have used successfully more than once.
2. Remove any real names, figures or confidential details.
3. Write it up using the format in the example, with yourself as author.
4. Share it with one colleague and ask them to try it on a made-up case.
5. Note what they changed and save it as version 1.1.

Going further

• Many assistants, including Claude, ChatGPT, Microsoft Copilot and Gemini, let you save instructions or create custom assistants that can be shared within an organisation; check which sharing options are approved.
• Store tested prompts as reusable commands or skills, so colleagues can run them in one step.
• Add a simple rating or "used this" counter to each entry to see what is popular.
• Review the library every quarter: retire old entries to an archive rather than deleting them.
• Pair each champion with a manager, so the freed time and lessons reach team planning.

Key words

• Prompt library: a shared collection of tested prompts and instructions.
• Skill: a saved set of instructions, examples and sometimes scripts that the assistant can reuse.
• AI champion: a colleague who helps others use AI well and looks after the shared library.
• Version: a numbered edition of a prompt, updated when it changes.
• Adoption: how widely and often people actually use what has been shared.

Examples and figures in this guide are illustrative. AI tools change often: check what your organisation's approved tools offer today. Use approved tools only, and a person checks and decides.

Prefer to print or share it? Download this guide as a PDF (237 KB), or the complete AI Field Guide (1.7 MB).