AI Team Building Activity: A Shared Prompt Challenge

A shared prompt challenge is one of the most practical AI team building activities available. Each person gives the same task to their own AI assistant, then the team compares the outputs, discusses what differed, and agrees on review habits and shared guidelines. The result is not just a conversation about AI — it is a set of concrete decisions your team can keep and apply.

A tool like fenbs keeps those decisions and the lessons from the exercise outside any one AI session, so the guidelines the team agrees on are available to every connected assistant going forward.

Why a prompt challenge works as a team exercise

Generic AI training sessions often cover what AI can theoretically do. A prompt challenge makes the team actually use AI on a real task, compare concrete results, and reach concrete agreements. The discussion is grounded in what people actually saw, not what the presenter described.

It also reveals how differently AI assistants handle the same task — and how differently team members prompt them. That variation is where the learning comes from. For ideas on wider AI adoption in your team, see AI adoption for your team.

Setting up the exercise

Choose a task that the whole team actually does: write a summary of a client meeting, draft a project update, suggest an agenda for a recurring call, review a short document for clarity. The task should take ten to fifteen minutes and produce something everyone can read.

Write the task brief in plain language. Give everyone the same brief so the outputs are comparable. Include enough context for someone who is not familiar with the specific project: background, purpose, and intended audience.

• The task and its context.
• The intended audience for the output.
• Any specific constraints: length, format, tone.
• What a good result looks like.

Running the activity

Give the team fifteen to twenty minutes to work on the task with their own AI assistant. Each person should try their natural prompting approach, not a standardized one. The variation is intentional and useful.

Ask everyone to save their output and the prompt they used. The prompt is as important as the output: two people can get different results from the same assistant by prompting differently.

Variations for different team sizes and remote groups

The activity works for in-person and remote teams. For a remote group, share the task and its context in a shared document before the session. Each person works individually for fifteen to twenty minutes, then pastes their output and the prompt they used into the shared document. During the call, the team reviews outputs together using screen sharing.

For a large team, divide into groups of three or four. Each group runs the activity independently, then shares one output and one key observation with the wider team. This keeps the discussion manageable while producing more diverse examples to compare.

Comparing outputs together

Share the outputs — without attribution first if that helps the discussion. Ask the team: which output would you actually use? What is missing from each? Where did the AI add something useful that was not in the task description? Where did it make something up or get something wrong?

Then share the prompts. Where a better output came from a more detailed prompt, make that visible. The team often finds that the prompts that work best are the ones that treat the AI like a briefed colleague: clear context, a specific task, and an explicit audience.

Turning results into team decisions

The exercise produces raw observations. The goal is to turn those observations into decisions: specific things the team agrees to always do, or never do, when using AI on shared work. Prioritize: if you have ten observations, ask which three would most improve the quality of AI work next month. Record those three first. A short list everyone applies is worth more than a long list nobody reads.

Examples: always include the intended audience in prompts for client-facing documents; always check factual claims against primary sources before sending; never paste AI output directly into a client email without a human review step. These are decisions, not preferences — they apply to everyone on the team.

How to do this with fenbs

1. Create a free account and create a team kanban board for your AI working group.
2. Add the exercise brief as a task in the board, with the context and constraints as notes.
3. Run the activity. After comparing outputs, facilitate a discussion and write down the decisions the team reaches.
4. Record those decisions as Decisions and rules in fenbs — standing rules apply to all future work; one-time decisions apply to the exercise.
5. Add a lessons learned entry for anything the exercise revealed that should change how the team prompts or reviews AI output.

For building a longer-term AI strategy for your team, see AI strategy for your team. For AI memory across team projects, AI memory for teams explains how shared context works in practice.