How does a team change when AI takes on part of its work?
Torsten Schollmayer
AI prepares a draft that someone else needs to use. For that handover to work, the team needs to agree who defines the task, checks the result and makes the next commitment.
The paper connects several stages. The image reflects the handovers within a team.
Imagine a team preparing answers to common customer questions. A colleague currently collects the questions, talks to the people responsible and writes a draft. The team now wants AI to organise the questions and prepare initial answers. It needs to consider how the colleague will work with that draft and what she will pass on to others.
Agree what you want AI to work on
Before anyone asks AI to do the work, the team needs a shared understanding of the result. Should it improve the wording of existing answers? Prepare a first draft from available documents? Suggest questions that have no agreed answer yet? Each task needs different preparation and checking.
In this example, the colleague could specify that AI should use only approved documents. Where a detail is missing, it should flag the unanswered question. She can then raise it with the people responsible. This agreement also helps the rest of the team understand what has been done and what still needs attention.
The person defining the task needs to know which documents are current and who can answer questions about the content. The team should account for that preparation when it allocates the work. Otherwise, it may expect a finished text from someone who still has to obtain the information needed to write it.
Who takes responsibility for each step?
01
You define the task.
You agree which questions to answer and which documents AI should use.
02
AI prepares the draft.
It organises the questions and drafts answers from those documents. Missing details remain visible.
03
You approve the answer.
You resolve outstanding questions with the people responsible and check whether the text is ready to use.
The example follows the preparation of a customer response. The team agrees who takes on each task.
Include unresolved questions in the handover
A well-written draft can look as though the work is finished. But the next person may not know which documents it draws on. They may also be unable to tell whether anyone has checked its claims. A handover therefore needs a short explanation of the work completed so far.
The colleague might explain that she has checked the answers about delivery times. For a question about returns, she is still waiting for information from the person responsible. Someone taking over can use the checked sections and resolve that specific question. They do not have to reconstruct the entire process.
This explanation need not be a lengthy report. It needs to give the next person enough information to understand their task. Which answers can they use? What still needs checking? Who should they speak to? The team can work through a real handover to find out which details it needs.
Make the commitment between people clear
An AI-generated draft is not a commitment to a colleague or customer. A person needs to say what they can deliver and when. Commitment-Based Management, or CBM, provides a useful basis for these agreements. One person makes a clear commitment to another. Both understand what it covers.
In this example, the colleague can commit to delivering the checked answers by an agreed date. If essential information is missing, she needs to be able to raise that before committing. Perhaps the team can add the outstanding answers later. Perhaps it needs a different date. The people relying on the commitment make that decision together.
Checking and follow-up questions are part of the task. If the team plans only for the finished text, it leaves that work out. A quickly produced first draft says little about when the colleague can hand over an answer that is ready to use.
Follow a complete task through the team
I would start by choosing a recurring task and following it through to completion with the team. How does the question reach them? Who obtains the information? Who checks the draft? What does the person taking over receive? That concrete process gives the team a basis for deciding what AI should do.
After the first handover, the team can look at where further explanations or questions were needed. Those observations inform its next agreements. The people involved can change how they work together while understanding what they remain responsible for.
