Using AI as a team: 5 rules for sharing best practices

In many teams, everyone uses AI on their own. One person has found the right prompt for summarising meetings, another has a trick for preparing quotes, and a third checks everything twice since being caught out. These discoveries each stay on one computer, and the team keeps retracing the same steps. Using AI as a team means, among other things, turning isolated experiments into shared practices. These five rules are a starting point.

What sharing brings

If six people each save twenty minutes on the same task, the total gain is two hours, before counting the time spent sharing and checking. A mistake spotted by one person saves the others from making it. Sharing increases the gains and reduces the risks.

It also reassures people. The most cautious get started more easily when they see a colleague using AI for a concrete task, and the most enthusiastic benefit from putting their discoveries in front of others.

Rule 1: a shared library of prompts

Create a single place to store the prompts that work. A simple shared document is enough. For each one, note what it’s for, the prompt itself and an example result, then sort them by task (emails, meeting notes, quotes).

A newcomer finds in five minutes what others took months to discover. To write effective prompts, use our method for writing a good prompt.

Rule 2: clear rules on what never gets shared

Define which data is off limits and which can be used in each approved tool. Review these rules when uses, tools or processing terms change. Passwords and access codes should never be typed into prompts. Also say which tools are allowed and for what.

One page is enough, as long as it’s simple and everyone knows it. Our article on what never to share with a chatbot can be your starting point.

Rule 3: check before you share

What AI produces is a first draft. Before a text or a figure leaves the team, someone rereads it. The higher the stakes, the more checking matters, and a quote or a customer document always deserves a second look.

Decide who checks what, and how. The most rigorous people in the team are often happy to take on this role. To know what to check, read how to spot AI hallucinations.

Rule 4: champions in the team

To get started on simple uses, you can rely on internal champions, with support suited to your needs: curious people who like testing and explaining, and who keep best practices circulating. A champion shows more than they train and adapts their advice to each person.

This role often falls to the Experimenter, who likes testing and sharing their finds. They can work with someone who likes formalising and maintaining practices, for example a participant who recognises themselves in the Pillar profile.

Rule 5: a regular check-in

Plan a short, regular moment, for example fifteen minutes once a month. Everyone shares a trick that works or a use they’ve dropped, then the library and, if needed, the rules get updated.

This check-in stops tools nobody uses from piling up and best practices from being forgotten. It also gives the quieter members of the team a chance to share what they’ve discovered.

A good AI team brings different profiles together

The teams that succeed with AI bring complementary profiles together. Trailblazers explore, Builders structure, Scouts move forward with care and Stewards maintain practices. Checking results remains a shared responsibility.

To explore your team’s tendencies and how its profiles complement each other, discover the team test, in preparation. No profile is required: the aim is to discuss which practices to strengthen. To start with yourself, take the free test. In about 3 minutes, you’ll discover what your profile can bring to a team and a complementary profile. For the manager’s view, read how to get the whole team involved in a small business.