You will build a one-page capability map of your work, tested on real tasks, that you can share with your manager or team.
At some point soon, someone will ask you how you use AI in your job. It might be your manager in a performance review, a team lead drafting a policy, or a recruiter in an interview. Most people answer with a vague list of tools. This project gives you something better to hand over: a one-page map of your actual work, showing what you give to AI, what you do together and what you keep, with evidence for each choice.
You will build a capability map: a table of fifteen to twenty recurring tasks from your role, each scored, placed in a zone and paired with the risk to watch. Three of the placements will have been tested on a real assistant. It should fit on one page and make sense to a colleague who has not taken this course.
Set aside about an hour. You need the capability map template below this lesson, your calendar or to-do list from the past two weeks, and an assistant. Keep confidential details out of the map itself. "Prepare monthly payroll summary" is fine, while the names and salaries are not.
Start with the list. Go through your calendar, sent emails and to-do list for the past two weeks and write down every task that comes up regularly. Aim for fifteen to twenty. Be specific: "write the weekly sales update for the regional team" is better than "emails". For each, estimate roughly how many hours it takes in a typical week. Rough is fine, and these numbers are what make the map useful to a manager.
Then score each task from one to five on the three questions from lesson 8.2: how quickly you can check the output, how cheap an error would be, and how well the task fits what models do well. In each case five means better suited to AI. Then place each task in a zone from lesson 8.3: hand over, work together or keep. Let the scores guide you, but use your judgement on accountability, confidentiality and relationships, and note your reason when you overrule the scores.
Scores are a prediction, and lesson 8.1 showed that the edge is hard to see until you test it. Pick three tasks from the hand over and work together zones, ideally ones where you were least sure of the placement.
Run each one on an assistant, using real but non-confidential material. Then check the output the way the zone requires: a light check for hand over, the full routine from lesson 6.3 for work together. Note what went well and what went wrong, and how long the check took.
Then adjust. If a hand over task needed heavy correction, move it to work together. If a work together task came back clean and quick to check, note that it might move after a few more tries. Mark these three tasks as tested on your map.
Add one more column, naming the main risk for each task in the first two zones. Draw on what the course covered. Invented figures or references come from module 6, and details lost in long files or long chats from module 5. Misread images, receipts or audio come from lessons 7.1 and 7.2, while wrong actions by an agent are from lesson 7.4. Outdated information after the training cutoff goes back to lesson 4.4. Naming the risk tells you, and anyone reading the map, exactly what to look for.
Here is part of one example. Li Ting is an executive assistant to a director at a property developer in Singapore. Her map has eighteen tasks. A few rows look like this:
Draft replies to routine meeting requests, 2 hours a week. Scores 5, 5, 5. Hand over. Risk to watch: wrong date or time format. Check: confirm date and time against the calendar. Summarise the director's board papers into a one-page brief, 3 hours a week. Scores 3, 2, 4. Work together. Risk: missed details in the middle of long papers. Tested: the first summary missed a figure on page 30, so she now gives it one section at a time. Book travel for the director's trips to Jakarta and Ho Chi Minh City, 1 hour a week. Scores 3, 2, 3. Work together, with an agent tool allowed to search but not to pay. Risk: agent errors in dates and names. Handle staff complaints that reach the director's office, 1 hour a week. Keep. Reason: confidential and about people.
At the bottom, a short note names the one change she will make first. In her case, she will use an assistant for meeting replies starting next week and review how many needed correcting after a month. She also wrote down how much time she expects to save, so she can compare it later with what actually happened.
A finished map has at least fifteen tasks, all with hours, scores, a zone and a risk where it applies. Three are marked as tested with what you found, and there is one clear note about where you will start. If a colleague can read it in two minutes and see how you plan to work, it is done.
Complete the capability map template with at least fifteen tasks, three tested placements and a short note on the one change you will make to your working week first.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).