AI At Work

Take an accountant. The title might suggest one clear activity, yet the working day could include collecting documents, entering data, checking figures, explaining results, answering questions, preparing reports, and helping a client decide what to do next. Those tasks don’t all have the same exposure to AI.

Data entry and document sorting are easier to automate than trust-based advice. Producing a first draft of a report is easier to automate than deciding what the numbers mean for a client’s business. The job is not one solid block. It’s a collection of activities with different levels of judgment, responsibility, and human contact.

The same applies to a plumber. Scheduling, quoting language, invoice descriptions, and customer follow-up may be supported by AI. Diagnosing an unusual leak in an old building still depends heavily on physical inspection and experience. A gym owner may use AI to create member updates or organize a retention campaign, but the quality of coaching still depends on observing people, adjusting workouts, and building trust.

Today, write down the main tasks in your role. Don’t describe them with broad labels such as “marketing” or “operations.” Be specific:

• Answering customer emails

• Preparing weekly sales figures

• Writing job estimates

• Checking stock levels

• Summarizing meetings

• Training new staff

Then mark each task as one of three types: repetitive, judgment-heavy, or relationship-based. Repetitive tasks deserve your first attention because they’re often the easiest place to gain time. Judgment-heavy and relationship-based tasks deserve attention for a different reason: they’re where your value may become more important as routine work gets faster.

You’re not trying to predict your entire career. You’re identifying where change is most likely to start.

Day 2: Notice What AI Is Actually Good At

AI is powerful, but it isn’t magic. It’s especially useful when the work involves language, patterns, structure, or a large amount of information.

It can turn rough notes into a readable draft. It can compare two documents and point out differences. It can group customer feedback into themes. It can suggest questions for a meeting, explain a complicated passage, or create several versions of a message for different audiences.

That doesn’t mean the result is automatically correct. AI can misunderstand context, invent details, miss an important exception, or sound confident while being wrong. Its strength is speed and range, not personal responsibility.

Think of AI as a fast junior assistant. Give it a clear task and useful material, then inspect the result before you use it. If you give a vague command like “make this better,” the output can be superficially polished but incorrect, If you ask it to turn these customer complaints into five themes, include the complaint text, and quote the evidence for each theme, you’ve given it a job it can handle more reliably.

Look at yesterday’s task list and choose one repetitive task that involves words, documents, or patterns. Ask yourself:

• Does this task follow a fairly repeatable process?

• Could a first draft or first pass save me time?

• Can I check the result before it reaches a customer, client, manager, or colleague?

If the answer is yes, you’ve found a likely AI-assisted task. Notice the wording: AI-assisted. The aim isn’t to hand over responsibility without checking. The aim is to stop spending your best attention on work that doesn’t require all of it.

Day 3: Understand Which Jobs Are Changing First

Jobs with a high amount of routine digital work are changing quickly. That includes many administrative, customer support, sales, marketing, finance, legal, research, and content roles.

A customer support representative may use AI to search internal information and draft replies. A sales representative may use it to summarize calls, prepare follow-up messages, and organize account notes. A researcher may use it to sort a large collection of documents before reviewing the important material in detail. A legal professional may use it to compare clauses or locate relevant passages, while still making the final judgment.

This pattern doesn’t mean every person in these roles will be replaced. It means the work inside the role is being rearranged. Fewer hours may go toward producing a first draft. More time may go toward checking quality, handling unusual cases, talking with customers, and making decisions.

Physical and hands-on jobs aren’t untouched, either. AI can support dispatching, quoting, stock planning, customer communication, and training. But jobs that depend on physical presence, changing environments, manual skill, or close human trust are harder to replace end to end.

The strongest protection isn’t a particular job title. It’s the ability to combine useful judgment with new tools. A bookkeeper who can explain financial choices clearly and use AI to handle routine preparation is in a stronger position than one who only performs the routine preparation. A recruiter who understands people, evaluates fit, and uses AI to organize candidate information has more leverage than one who spends most of the day copying details between systems.

Review your own role today. Which parts are already being discussed in your industry as “automated,” “assisted,” or “streamlined”? Don’t react to the label. Write down the actual task involved. That’s where the practical answer is.

Day 4: The Difference Between Replacing a Task and Replacing a Job

A task is one activity. A job is the full set of responsibilities around that activity.

AI might replace the task of creating a meeting summary. It doesn’t necessarily replace the person who knows which decisions matter, notices that two departments disagree, follows up with the right people, and makes sure the work gets done.

It might replace the first draft of a product description. It doesn’t necessarily replace the person who understands the customer, chooses the right position in the market, checks whether the claims are accurate, and decides which version should be published.

This distinction matters because headlines often collapse the two. “AI will replace writers” may really mean that fewer people will be paid only to produce basic first drafts. “AI will replace analysts” may really mean that routine data preparation will require less manual work. “AI will replace assistants” may really mean that scheduling, sorting, and standard communication will take less time.

Some jobs will shrink. Some will change significantly. A few may disappear or become much smaller. New responsibilities will also appear: checking AI output, managing information, setting standards, protecting confidential data, and making decisions when the situation doesn’t fit the pattern.

Take one task from your list and complete this sentence:

“AI could handle the first pass of ______, but a person still needs to ______.”

For a property manager, that might be: “AI could handle the first pass of tenant email replies, but a person still needs to decide how to handle disputes and urgent safety issues.” For a consultant, it might be: “AI could handle the first pass of research notes, but a person still needs to test the conclusions against the client’s situation.”

That sentence gives you a more useful view than either panic or denial.

Day 5: Find the Human Parts That Gain Value

When routine work becomes cheaper and faster, human judgment often becomes more visible.

Customers still want someone who understands their situation. Employees still need clear decisions. Clients still pay for confidence when the answer isn’t obvious. Teams still need people who can spot a problem early, ask a useful question, and take responsibility for a result.

AI can produce ten options. Someone still has to choose the option that fits. It can summarize a complaint. Someone still has to decide whether the complaint reveals a serious service problem. It can identify patterns in sales data. Someone still has to decide what action the business can afford and what risk it should accept.

This is why learning AI isn’t only about becoming faster. It’s about moving your effort toward work that benefits from context, taste, accountability, communication, and judgment.

Look at the task you chose yesterday. Write down the human contribution that remains. It might be checking accuracy, understanding the customer, making a difficult call, or taking responsibility for the outcome. Then ask whether you’re spending enough time developing that contribution.

If AI handles more of the preparation, your advantage may come from being better at the decision that follows.

Day 6: Watch the Quiet Changes Around You

The clearest signs of change may not come from a company-wide announcement. They may appear in small adjustments.

A manager asks for faster reports. A colleague starts producing more proposals without working longer hours. A business stops hiring for a role that used to handle basic document preparation. A customer begins expecting an answer within minutes instead of a day. A team adopts a new process because one person found a faster way to complete it.

These changes can feel unfair if you only see the result. Pay attention to the process. What used to take three hours may now take forty minutes. The organization may not need fewer people immediately, but it may expect more output from the same team. That changes what good performance looks like.

For one week, keep a simple change log. Each day, record one example of AI changing a task, expectation, or workflow in your workplace or industry. Don’t collect rumors. Look for observable details: a new tool, a changed turnaround time, a different hiring requirement, or a task that no longer receives the same attention.

At the end of the week, look for a pattern. Is the change mainly about speed? Volume? Lower cost? Better service? Fewer errors? The answer tells you what employers and customers are beginning to value.

Day 7: Choose Your Next Move

You don’t need a complete AI strategy to stay useful. You need one sensible place to begin.

Choose a task that is repetitive, important enough to matter, and easy enough to check. Decide what a good result looks like before you use AI. Keep sensitive information out unless your organization has approved the tool and process. Review the output. Measure the time saved and the corrections required.

Then make a second decision: what will you do with the time you recover? If the answer is “more work,” you may simply create a faster treadmill. Use some of the capacity for customer conversations, deeper analysis, better preparation, training, or work you’ve been postponing because routine tasks consumed the day.

AI is changing jobs by changing their ingredients. The routine parts are becoming easier to produce. The parts involving judgment, trust, context, and responsibility are becoming more important to identify and strengthen.

The real divide won’t be between people who use AI perfectly and people who don’t. It will be between people who examine their work and adapt, and people who wait for change to be explained to them after it has already affected their value.

Start with one task. Separate it from the job around it. Then decide where your human contribution matters most.

Day 7: Choose Your Next Move, Action Plan,

Pick one repetitive, important, and checkable task. Define success before you start: time saved goal, acceptable error rate, and where the output must be reviewed.

Implementation steps:

• Select the tool and confirm it’s approved for any sensitive data.

• Create a one-line prompt and a short verification checklist.

• Run two trials, record time and corrections.

• Reallocate 25, 50% of saved time to one higher-value activity for two weeks.

• Evaluate: did decisions improve? Did customer responses change?

If the answer is “more work,” stop. Use the time to improve judgment-based tasks instead. Small experiments beat big plans here.

Day 7: Choose Your Next Move, Action Plan,

Pick one repetitive, important, and checkable task. Define success before you start: time saved goal, acceptable error rate, and where the output must be reviewed.

Implementation steps:

• Select the tool and confirm it’s approved for any sensitive data.

• Create a one-line prompt and a short verification checklist.

• Run two trials, record time and corrections.

• Reallocate 25, 50% of saved time to one higher-value activity for two weeks.

• Evaluate: did decisions improve? Did customer responses change?

If the answer is “more work,” stop. Use the time to improve judgment-based tasks instead. Small experiments beat big plans here.

Non-office example: A Plumber’s Shift

AI doesn’t touch the wrench or the pipes, but it can change the working day, The routine parts, scheduling, quoting, follow-up messages, parts ordering, can be automated or handled faster. That frees you to spend more time diagnosing strange problems, teaching apprentices, or building repeat customers through better service.

Try this micro-test: automate quote messages and parts lists for three jobs. Track time saved per job and one quality measure (first-time fix rate). If quotes improve and you fix more issues on the first visit, you’ve increased real value.

Day 4: Replacing a Task vs. Replacing a Job (Compressed)

A task can be automated; a job bundles tasks and responsibilities. AI may do the first draft of a report, but it won’t own the responsibility for decisions the report supports.

Complete this sentence for one task: “AI could handle the first pass of ____, but a person still needs to ____.” Use the result to define your human role: checking accuracy, making the call, handling exceptions, or building trust.

Quick decision guide:

• If the task is repeatable and easily checked, automate the first pass.

• If errors would cause harm or require legal judgment, keep human control.

• If the task creates customer-facing trust, protect the human touch.

Record one change to your weekly schedule: allocate the time saved to a judgment task (client calls, strategy, coaching) for at least two weeks. Observe whether outcomes improve.

Day 2: Notice What AI Is Actually Good At

AI excels at language, patterns, and scale: turning notes into drafts, grouping feedback, comparing documents, creating versions for different audiences. Treat it like a fast junior: give clear material, a precise instruction, and a way to check the result.

Mini checklist before you ask an AI:

• Provide source material (notes, emails, transcripts).

• Specify the output format and length.

• Ask for evidence or citations for key claims.

• State what not to change (tone, legal terms, numbers).

Try this: pick the repetitive task from Day 1. Run one short experiment, record time saved and the number of corrections needed. Repeat once. Use the results to decide whether to adopt the AI step permanently.

Day 1: Separate the Job From the Tasks

A job title hides the work inside it. Break the title into specific tasks you actually do: not “marketing” but “answering customer emails,” “writing weekly reports,” “preparing client invoices.” Write five to eight items.

Then mark each task with one of three labels:

• Repetitive, follows a set pattern and is often time-consuming.

• Judgment-heavy, needs domain knowledge or decision-making.

• Relationship-based, depends on trust, nuance, or empathy.

Quick task-audit (5 minutes):

• Pick one task. Note the inputs (documents, data, people) and the usual steps you take.

• Ask: “Could a first draft or first pass save me time?” Yes/No.

• Ask: “Can I check the result before it reaches a customer or manager?” Yes/No.

If both answers are Yes, this is a prime candidate for AI assistance.

What success looks like: measure time now, then after two uses of AI. If you save at least 25% of time and still make fewer than two substantive corrections per output, you’ve found useful leverage.

End of chapter one. 6 more chapters and about 12,826 words in the full book.