The first time many professionals open an AI chatbot at work, they encounter an oddly uncomfortable problem.

They know AI is supposed to save time.

They've heard colleagues talk about prompts.

They've watched someone generate an impressive report in seconds.

They've probably read that AI can write, summarize, analyze, research, brainstorm, and automate.

Then they open the chat box and think:

"Okay… what am I actually supposed to ask it to do?"

That's the part nobody explains.

The problem usually isn't that AI can't help with your job. Research published in 2026 shows generative AI is already being used across a broad range of occupations and tasks, including routine information work.

The harder problem is recognizing which parts of your existing workload are worth handing to AI.

So instead of giving you another list of trendy AI tools, let's start with something more useful:

Your to-do list.

Start With Your To-Do List, Not an AI Tool

If you don't know what to ask AI, don't open an AI app and wait for inspiration.

Open your calendar or task list instead.

Look at the work you already need to complete this week.

Then look for tasks involving:

• Writing from scratch

• Rewriting or editing

• Summarizing information

• Organizing messy notes

• Repeating the same process

• Researching a question

• Preparing for a meeting

• Brainstorming possibilities

• Turning one format of information into another

Those are potential AI use cases hiding in plain sight.

For example, suppose you're a manager and tomorrow you have a meeting with your team.

"Prepare for the meeting" is too vague to hand to AI.

But you could ask AI to:

• Turn previous meeting notes into a briefing

• Create an agenda

• Identify unresolved issues

• Generate questions you should ask

• Role-play difficult questions employees might raise

• Turn the final notes into a follow-up summary

Or suppose you're in sales.

Instead of asking AI to "help with sales," you might ask it to turn customer-call notes into a concise CRM summary and identify the three most important follow-up actions.

The task comes first.

AI comes second.

Your existing workload is actually your AI use-case library.

Learn the Six Things AI Is Especially Good At

You don't need to memorize hundreds of AI applications.

A simple mental model is enough.

Most useful workplace requests fall into a handful of categories.

1. Draft

Give AI the first version of something you need to write.

Emails, reports, proposals, agendas, announcements, outlines, and internal communications are obvious examples.

2. Transform

Take something you already have and change it.

Make it shorter.

Make it clearer.

Change the tone.

Turn paragraphs into bullet points.

Turn notes into a report.

Turn a technical explanation into something a customer can understand.

3. Summarize

Give AI information that is too long to process comfortably and ask it to extract what matters.

Meeting notes, reports, research, email threads, customer feedback, and documents are common examples.

4. Organize

Give AI messy information and ask it to structure it.

Create categories.

Extract action items.

Identify deadlines.

Create a checklist.

Group similar feedback.

Turn scattered notes into a logical outline.

5. Brainstorm

Use AI when you need possibilities rather than a final answer.

Ask for ideas, alternatives, objections, risks, questions, or different approaches.

6. Analyze

Ask AI to compare information, identify patterns, challenge assumptions, surface gaps, or help you think through a problem.

These categories are useful because they turn a vague question like "How can I use AI?" into something much more concrete.

When you encounter a task, ask:

Am I trying to draft, transform, summarize, organize, brainstorm, or analyze something?

If the answer is yes, there's a good chance AI can at least help with part of it.

Stop Saying "Do My Job"

One of the fastest ways to get disappointing AI responses is to give it a vague assignment.

Consider:

"Help me with this report."

AI has almost no idea what "help" means.

Now compare it with:

"Turn these project notes into a weekly report for my manager. Separate completed work, current priorities, blockers, upcoming deadlines, and decisions that require attention. Keep it concise and don't invent information that isn't included in my notes."

That's a completely different request.

A useful prompting framework is:

Context + Task + Audience + Constraints + Output

You don't have to use those labels every time.

But give AI enough information to understand:

What is happening?

What do I want done?

Who is this for?

What should it avoid?

What should the result look like?

For example, an HR professional could say:

"Turn these interview notes into a structured candidate summary. Separate factual observations from interpretation, highlight evidence related to the role requirements, and identify areas that may need follow-up. Don't invent information."

A manager could say:

"Review these team updates and identify common blockers, overdue items, and decisions I need to make. Group similar issues together and prioritize them by urgency."

A sales professional could say:

"Turn these customer-call notes into a concise CRM summary. Identify the customer's main problem, objections, buying signals, next step, and information I still need."

Notice what these prompts have in common.

They're not magical.

They're specific.

Prompting isn't about discovering secret words that make AI suddenly become intelligent.

It's about explaining the job clearly.

Ask AI to Make the Work Smaller, Not Just Faster

Here's an even more useful question to ask:

"Which parts of this task don't require my full attention?"

That's often better than asking:

"Can AI do this entire task?"

You don't necessarily want AI to take over your work.

You want it to reduce the amount of low-value effort surrounding the part where your judgment matters.

Think about the division like this:

AI drafts → you approve.

AI summarizes → you verify.

AI organizes → you decide.

AI brainstorms → you select.

AI identifies patterns → you investigate.

AI creates options → you choose.

That distinction matters.

Research into organizational decision-making increasingly emphasizes that AI can support different stages of work while human intervention remains important, particularly when judgment and choice are involved.

This is also a much less intimidating way to introduce AI into your work.

You don't have to surrender control.

You can simply remove some of the work that happens before your expertise is needed.

Turn Your Best AI Requests Into Workflows

Suppose you discover that AI does an excellent job turning your meeting notes into a follow-up summary.

Don't start from zero next week.

Save the prompt.

Then make it reusable.

Create a template such as:

Meeting purpose: [insert]

Attendees: [insert]

Notes: [insert]

Then have AI consistently produce:

• Key decisions

• Action items

• Responsible people

• Deadlines

• Unresolved questions

• Follow-up communication

Now you're no longer just "trying AI."

You've created a workflow.

Use this progression:

Notice → Ask → Review → Improve → Save → Repeat

Notice a repetitive task.

Ask AI to help.

Review the result.

Improve your instructions.

Save what worked.

Repeat it next time.

Eventually, you can explore deeper automation—but you don't need complicated integrations to get your first win.

In fact, research on workplace AI consistently points toward structured, clearly defined tasks as an important area of practical value.

The goal isn't to collect 500 prompts.

It's to build a small collection of prompts that reliably solve your recurring problems.

When You Don't Know What to Ask AI, Look at What You're Already Doing

If you open an AI chatbot and don't know what to type, don't assume you're bad at AI.

Look at tomorrow's workload.

Ask:

What am I writing?

What am I summarizing?

What am I repeatedly organizing?

What am I preparing?

What am I researching?

What am I doing manually that follows the same pattern every time?

Those questions will reveal potential AI applications inside an ordinary workday.

And you don't need to start with your most complicated task.

Start with the annoying one.

The 15-minute task.

The repetitive email.

The weekly report.

The meeting notes.

The document that takes forever to turn into something useful.

Give AI a clearly defined role, review the result, and improve the process.

That's how AI stops being something you're vaguely "supposed to learn" and becomes something you actually use.

If you want a more structured way to make that transition, AI in a Day: The Non-Technical Professional's Playbook for Automating Work, Looking Smarter is designed for exactly this problem. It's a 25-page practical playbook for non-technical professionals who don't want to spend weeks exploring AI—they want to use it on real work and feel the difference immediately.

You're not bad at AI. You may simply not have had the right system yet.

Get AI in a Day here.