At work, AI is often most useful for the first pass. It can turn scattered notes into a list, compare options you provide, or draft a message. Decisions, fact-checking, and responsibility remain with you. When you divide a task into parts before opening an assistant, you know which part you are giving to AI and why.
Which parts of work are suitable for AI to prepare
- Suitable: a draft based on information you provide, structure for scattered notes, a comparison of options you provide, and a checklist for a familiar work process.
- Not suitable as-is: a decision that affects a customer or employee, information you cannot verify, or material you are not allowed to give to the service you use.
Divide the task in three steps
- 1
Break the task into parts
List the stages of the task: what you read, organise, decide, and write.
- 2
Choose a part for AI to prepare
Mark the part where a draft, structure, or comparison would help. Write one sentence explaining why AI is suitable for that part and how you will check the result.
- 3
Keep judgement, responsibility, and confidential information with you
Name the parts you will do yourself and why: a decision affects people, information needs checking, or material must not be shared.
Example: a new employee’s first day
The workplace in this example is fictional. You are preparing a schedule for a new employee’s first day and break the task into four parts: the schedule, a checklist, requesting system login credentials, and choosing who will onboard them.
- AI prepares a checklist from the schedule you provide, because the list follows a recurring structure and you can check it in one read-through.
- You put together the schedule yourself because you know when the team and facilities are available.
- You order access yourself because the request would contain the employee’s personal information.
- The manager chooses who will onboard the new employee because the decision affects colleagues’ working hours.
Request for the part AI will prepare
Create a first-day checklist for a fictional new employee. The workday lasts from 9 a.m. to 4 p.m. In the morning, the employee meets the team; at 11 a.m., they set up their work tools; and at 1 p.m., they go through one general stage of a customer process. Put the list in time order and mark separately any points the onboarding buddy still needs to clarify. Do not invent system names or customer information.
Try the same division when comparing options
When a team is choosing what to do next, AI can make the differences between options visible. The team makes the choice.
Comparison to support a decision
Organise these fictional team ideas for comparison: redesign the website homepage, send a monthly newsletter, and run a short customer survey. For each one, give the goal, estimated workload as either small or medium, the decision needed, and the next step. Do not choose a winner; make the differences visible to support a decision.
Check the division before you begin
- Does every part have an owner and a reason?
- Have decisions that affect people or money stayed with a person?
- Do you know how you will check the part AI prepares before using it?
At work, use a service your workplace has approved and follow its instructions about what material you may provide.
Free practice on this website
Break complaint handling into parts
The material describes a fictional work task. Answer the questions using the material.
Your answers stay only in this browser tab. They are not sent to AI Licenses’ server or to analytics.
Question 1/3
Question 1/3
For each of the five parts, mark whether AI would prepare it or you would do it yourself. Briefly explain each choice.
Skills to practice
Choose a task to delegate to AI
Concepts
Guides that use this
- Write a team status update without invented factsTurn a week of notes into a team update where every name, time and decision comes from the notes and missing information stays visible.
- AI can help you think through a decision. You remain responsible for it.Learn to spot automation bias and missing information, and keep human responsibility clear when using AI to support decisions.
- Write an AI request that gives you a useful draftWrite a request with a task, context, format and a checkable constraint, then improve the response with one follow-up request.
Sources and review
- Reviewed:
- September 11, 2026