Read the narration transcript
Welcome to lesson five, Using AI in your job. You now know how these tools work, how to ask, how to check, and what to keep out. Now, where should you actually start? This lesson helps you pick the right tasks, walks through examples by role, and covers how to be open with colleagues about AI's role in your work.
The best first uses of AI aren't your highest-stakes work. They're the tasks that eat time, involve a lot of text, and wouldn't cause harm if a draft needed fixing. Look for work that's repetitive, meaning you do it weekly or daily in roughly the same way. Work that's text-heavy, like writing, summarizing, reformatting or explaining. And work that's low-risk, where a mistake would be caught in review and wouldn't reach a customer, a regulator or a paycheck unchecked.
Here are some starting points by role. In each case, the AI produces a draft and you review it. In finance and accounting, it can explain a variance in plain language for a non-finance audience, draft a reminder about month-end deadlines, or write or fix a spreadsheet formula that you then test. In HR, it can draft a job posting from a list of duties, turn a policy into an employee-friendly FAQ, or prepare structured interview questions. In operations, it can turn rough notes into a step-by-step procedure, draft a vendor inquiry, or summarize a long process document into a one-page checklist. In sales and customer service, it can draft replies to common questions, adjust the tone of a response, or summarize a product sheet for a specific type of customer. Managers can use it to draft an agenda or structure feedback for a one-on-one. And for administrative work, it can draft scheduling emails, clean up meeting minutes, or reformat a list into a table.
A simple way to choose is to ask two questions about any task. How much time or effort would AI save? And how much harm could an error cause before someone caught it? High-value, low-risk work is where to start. Think routine emails, summaries of your own notes, and reformatting. High-value, higher-risk work can still work, but it needs a careful check step and possibly a second reviewer. A summary of contract terms, or a budget narrative, would fall here. Low-value tasks often aren't worth the effort of prompting and checking. And anything where a mistake could seriously harm a person, such as a hiring or disciplinary decision, needs human judgment at the center, with AI at most helping you organize your own thinking. Your organization may also have rules about specific uses. Follow those first.
Below this section is an interactive map. Choose a role, then pick a task to see where it lands. The map has four zones. Start here means high value and low risk. Add controls first means it's worth doing, but you need review or an approved tool in place before you rely on it. Optional means low value. And people decide means AI can help prepare, but a person makes and owns the decision. Pause the audio here and try a few roles. In finance and accounting, drafting variance commentary lands in start here, while posting journal entries lands in people decide. In HR, drafting a job posting is a good place to start, and deciding who to interview stays with a person. Treat each placement as a starting point for discussion. Your policy and the task decide the real answer.
When AI helped with something meaningful, say so. A short note works. Something like, I used our AI assistant to draft this and checked the figures against the source. That builds trust, helps colleagues judge how to review the work, and spreads good practice across the team. Talk with your manager about where AI fits in your role, and whether your team has expectations about disclosure. Some teams like to share prompts that work. Others set rules for client-facing material. Either way, it's better to have that conversation early.
The worked example turns an HR policy into an FAQ. It asks for six to eight questions employees are likely to ask, each with a short, plain-language answer based only on the policy text, plus a separate list of likely questions the policy doesn't cover. That's high value and fairly low risk. The policy is internal, it isn't personal data, and the output gets reviewed before anyone sees it. The not covered list is useful on its own, since it shows gaps HR may want to address. For the Try it exercise, Map your tasks, list five tasks you do at least monthly. Mark whether each one is repetitive, text-heavy and low-risk. Rate each one's value and risk as high or low, and pick the one that's high value and low risk. Then, using only non-sensitive information, try a first request for it in your approved assistant, and note how much editing the result needs. Hold on to that task. It's a good candidate for the routine you'll build in lesson six.
Here are the takeaways. Start with repetitive, text-heavy, low-risk tasks. Weigh the time saved against the harm an error could cause. And tell your manager and colleagues when AI helped, and how you checked it. Take the quiz below, then join us for the final lesson, Your first workflow.
Start with the right tasks
The best first uses of AI are not your highest-stakes work. They are the tasks that eat time, involve a lot of text and would not cause harm if a draft needed fixing. Look for work that is:
- Repetitive: you do it weekly or daily, in roughly the same way.
- Text-heavy: writing, summarizing, reformatting or explaining.
- Low-risk: a mistake would be caught in review and would not reach a customer, a regulator or a paycheck unchecked.
Examples by role
These are starting points. In each case, the AI produces a draft and you review it.
- Finance and accounting: explain a variance in plain language for a non-finance audience, draft a reminder to departments about month-end deadlines, write or fix a spreadsheet formula and then test it.
- HR: draft a job posting from a list of duties, turn a policy into an employee-friendly FAQ, prepare structured interview questions for a role.
- Operations: turn rough notes into a step-by-step procedure, draft a vendor inquiry, summarize a long process document into a one-page checklist.
- Sales and customer service: draft replies to common questions, adjust the tone of a response, summarize a product sheet for a specific type of customer.
- Managers: draft a meeting agenda, structure feedback for a one-on-one, turn a project update into a short note for leadership.
- Administrative work: draft scheduling emails, clean up meeting minutes, reformat a list into a table.
Weigh value against risk
A simple way to choose is to ask two questions about any task. How much time or effort would AI save? And how much harm could an error cause before someone caught it?
High-value, low-risk work is where to start: drafting routine emails, summarizing your own notes, reformatting. High-value, higher-risk work can still work, but it needs a careful check step and possibly a second reviewer: a summary of contract terms, a budget narrative. Low-value tasks are often not worth the effort of prompting and checking. And anything where a mistake could seriously harm a person, such as a hiring or disciplinary decision, needs human judgment at the center, with AI at most helping you organize your own thinking.
Your organization may also have rules about specific uses. Follow those first.
Loading the interactive example…
Be open about it
When AI helped with something meaningful, say so. A short note works: "I used our AI assistant to draft this and checked the figures against the source." This builds trust, helps colleagues judge how to review the work and spreads good practice across the team.
Talk with your manager about where AI fits in your role and whether your team has expectations about disclosure. Some teams like to share prompts that work. Others set rules for client-facing material. Either way, it is better to have the conversation early.
An HR policy into an FAQ
This is high value and fairly low risk: the policy is internal, not personal data, and the output gets reviewed before anyone sees it. The "not covered" list is useful on its own, since it shows gaps that HR may want to address in the policy.
Map your tasks
- List five tasks you do at least monthly.
- For each one, mark whether it is repetitive, text-heavy and low-risk.
- Rate each task's value (time saved) and risk (harm if an error slipped through) as high or low.
- Pick the one that is high value and low risk.
- Using only non-sensitive information, try a first request for that task in your approved assistant, and note how much editing the result needs.
How you’ll know it worked: You should end up with one specific task you can reasonably hand to AI for a first draft, and a sense of how much review it needs. That task is a good candidate for the routine you will build in Lesson 6.
Use only non-sensitive information, in a tool your organization allows.
Key takeaways
- Start with repetitive, text-heavy, low-risk tasks.
- Weigh the time saved against the harm an error could cause.
- Tell your manager and colleagues when AI helped, and how you checked it.





