You will produce a scored task audit and choose the tasks you will work on through this course.
The list of tasks you made in lesson 1.1 came from memory of one week, and a single week can be unusual. A quarterly review, a colleague on leave or a client crisis can push ordinary work off the list. To choose which tasks to work on with AI for the rest of the course, you need a structured audit of a typical week. Build it from your calendar, sent email and recent files, because they show what you actually did rather than what you think you did.
Set aside about thirty minutes and open your calendar, your sent email and your recent files. Start a spreadsheet with these columns:
Task Frequency per week Average minutes each time Minutes per week Checkability score Error-cost score Data flag
Fill in the task column first. Look back over the last two or three weeks of your calendar, sent folder and documents. Write down everything that repeats until you have at least twelve tasks. Your ten tasks from lesson 1.1 are a starting point, but check each one against the records.
Be specific. A broad task such as "email" needs to be split into separate rows, for example routine supplier replies, updates to your manager and chasers for late approvals. Each of these behaves differently with AI, so each needs its own row and its own scores.
For each task, enter how often you do it in a week and the average minutes it takes each time. Monthly tasks get a fraction: once a month is about 0.25 per week. Minutes per week is the frequency multiplied by the average minutes.
This column shows where your hours actually go, and it often surprises people. Small tasks done many times a day add up faster than the big task you dread.
Both scores run from 1 to 5. The checkability score is how easy it is to tell whether the output is right. A short reply to a routine query scores 5, because you can check it in a minute by reading it. A line-by-line reconciliation scores 1, because the only way to check it is to redo it.
The error-cost score is how much damage a mistake would do if it got through. An internal note, where a clumsy sentence costs nothing, scores 1. A figure in a client proposal, or a report management uses to decide budgets, scores 5.
Score quickly and trust your judgement. The scores are for sorting the list rather than for precision, and you'll refine them when you start testing.
The last column is the data flag. Mark any task that involves personal or confidential information: names with contact details, salaries, NRIC numbers, health information, client contracts, unreleased financial figures or pricing. Don't decide yet whether AI may be used on these tasks. For now the flag only keeps them separate until module 2, which covers what you may paste into AI, what to redact and how to check your company's rules.
A target is a task you'll work on for the rest of the course. It must pass three tests at once. It takes a meaningful amount of time each week, it scores checkability 4 or 5, and it scores error cost 1 or 2. A flagged task can become a target after module 2, but if you have enough unflagged candidates, start with those.
Sort the sheet by minutes per week. Read down from the top and highlight the tasks that pass all three tests, up to a maximum of five. If fewer than three pass, relax the error-cost test to 3 for one task and plan to check that task more carefully.
Priya handles suppliers, reports and staff rosters, and she reports to a regional manager. All her figures here are example figures.
She sends routine supplier replies about 40 times a week at 6 minutes each, which comes to 240 minutes a week. She scored them checkability 5 and error cost 2, with no data flag, and they went straight onto her target list.
Her Monday report runs once a week and takes 120 minutes. It scored checkability 3, because the figures have to be checked against three source sheets, and error cost 4, because management uses it. It did not pass the tests, so she marked it for module 5, where a formula with a check cell might make a report like this safer.
The notes after her weekly operations meeting take 25 minutes once a week. She was in the meeting and can spot errors quickly, so she scored them checkability 4 and error cost 2, which made them a target.
Her monthly supplier invoice reconciliation takes 90 minutes, or about 22 minutes a week. It scored checkability 1 and error cost 5, and she flagged it for confidential pricing, so it was not a target. Drafting staff rosters takes her 45 minutes a week and involves names and leave details. She flagged that too and set it aside until module 2.
After twelve rows she had three clear targets: supplier replies, meeting notes, and a monthly update to her regional manager, which scored checkability 4 and error cost 2. She then chose to add the Monday report as a fourth target. That goes beyond the fallback rule, which relaxes error cost to 3 for one task only when fewer than three pass. She already had three, and the report's scores fall outside even the relaxed test. She added it knowing it would need more careful checks than the other three. If you make a choice like that, note the reason in your sheet so anyone reading it can follow. Otherwise, choose your targets by the three tests and the fallback rule.
Her baseline log from lesson 1.3 already covered three of the four targets, so she added the monthly update to it the same day. Your own targets should all be in your baseline log as well.
A finished audit has at least twelve rows. Each row has weekly minutes, two scores and a data flag where one applies, and three to five rows are highlighted as targets. Someone reading the sheet should be able to see why you chose those targets and not the others.
Now build your own audit with the steps above. Work from your real calendar and sent folder, because memory flatters some tasks and hides others.
Complete the task audit sheet with at least twelve tasks, scores and data flags, and highlight your three to five target tasks.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).