Your first AI teammate needs a clear job

Think about a task that returned to your desk this week. Perhaps a meeting needed a summary, an inquiry needed a thoughtful response, or scattered notes needed to become a campaign brief. That recurring job is a useful starting point for an AI teammate.

Here, an AI teammate means an assistant with a defined task and a person responsible for its work. It could begin as a saved brief and a repeatable conversation. Connections to other tools can come later, when the process is clear enough to test.

Choose a job you understand

Start with work you can judge. You need to recognize a useful result, spot an omission, and know when to ask for another attempt. Choose a task that appears regularly and has a manageable set of inputs. Keep the first version small enough to inspect from beginning to end.

For a fictional marketing team, the task might be turning approved discovery notes into the first draft of a campaign brief. The assistant can organize the information, identify unanswered questions, and suggest a structure. The campaign owner still decides what the team should promise and produce.

Write a short job description

Give the assistant five things: the purpose of the task, the material it may use, the format you need, the checks the result must pass, and the point where a person takes over. Include an instruction for missing information so a confident sentence does not quietly become an invented fact.

Fictional practice brief: Use these sample discovery notes to draft a one-page campaign brief. Include the audience, objective, offer, message, proposed deliverables, and open questions. Mark missing information. Keep proposed ideas separate from decisions in the notes. Return a draft for review; do not publish or send it.

Save that brief beside a good example and a short review checklist. Together, they explain the job more clearly than a collection of disconnected prompts.

Inspect the work before expanding it

Compare the draft with the original notes. Are the audience and objective accurate? Did the assistant turn a possibility into a commitment? Is the next action clear? Make corrections and record why they were needed. Those observations can improve the next instruction.

Then try a second example with something missing or ambiguous. A process that works on one tidy input still needs testing on the ordinary variations your team encounters. Count preparation, review, and correction time when deciding whether it is useful.

Leave someone in charge

Write down who supplies the input, who reviews the output, and what happens when a required detail is absent. Keep a manual way to finish the task. If the workflow will send messages or change records, agree on permissions and approval steps before adding those actions.

In Waymaker Agency's AI Teammate Workshop, the focus is practical team learning around a useful work task. Custom team bookings are available. If you would prefer a public cohort, you can express interest; a public date and price have not been announced. Bring a recurring task and a clear picture of what good work looks like.

Explore the AI Teammate Workshop