How to Make AI Work for You Instead of the Other Way Around

An Accounting Against the Machine Side Quest

We’re taking a brief detour from ProAdvisor certification.

Don’t worry. There will be plenty more QuickBooks.

But if you’re learning modern accounting software, you’re going to run into AI whether you particularly asked for it or not. It’s being built into software, search tools, productivity platforms, browsers, and damn near everything else.

So I think learning how to use it intelligently belongs right beside learning the software itself.

And no, the lesson isn’t:

Ask ChatGPT to do your work.

Quite the opposite.

I want to show you how I turn AI into a tool that works around my brain, my projects, my learning style, and my goals—while keeping myself firmly in charge of the operation.

Because typing a question into an AI chatbot is easy.

Teaching one how to be genuinely useful to you takes a little more work.

First: AI Is Not the Smart Kid You Copy Homework From

This is probably the most important thing in this entire post.

AI can be wrong.

Confidently wrong.

Beautifully formatted, grammatically impeccable, sounds-like-it-absolutely-knows-what-it’s-talking-about wrong.

I’ve had AI help me tremendously.

I’ve also asked it to make my lower lip slightly fuller in a photograph and received an entirely different, considerably younger woman.

We have history.

So I don’t treat AI like an authority.

I treat it like an assistant.

There’s a massive difference.

An assistant can research, organize, explain, brainstorm, quiz me, help me find gaps in my understanding, and turn my chaotic notes into something usable.

But I’m still responsible for the final product.

Especially when we’re talking about accounting information that could affect somebody’s actual business.

Step 1: Stop Talking to It Like Google

Here’s a perfectly reasonable AI prompt:

Explain bank reconciliation.

You’ll probably get a perfectly reasonable explanation.

And then you’ll forget half of it.

Instead, give the AI a job.

Try:

I’m studying QBO for ProAdvisor certification. Teach me bank reconciliation like you’re an experienced trainer sitting beside me. Start by explaining why businesses reconcile accounts before showing me the QBO workflow. Use a realistic small-business example. Keep the language conversational and stop periodically to check whether I understand.

Now we’ve given it:

Context: I’m studying QBO.

Role: You’re helping teach me.

Goal: I need to understand reconciliation.

Method: Explain the concept before the software workflow.

Style: Conversational.

Interaction: Check my understanding instead of dumping information on me.

Same AI.

Much better assignment.

Step 2: Tell It Who It Is

This sounds sillier than it actually is.

You don’t have to give your AI a name and develop an entire ongoing relationship with it.

I did.

That’s between Roland and me.

But assigning a role is enormously useful.

If I’m studying, I might want:

Act as an experienced QBO trainer helping a new support professional understand accounting concepts and the software workflows behind them.

If I’m troubleshooting:

Act as my troubleshooting partner. Don’t immediately give me a solution. Help me identify what we know, what we don’t know, and what we should verify next.

If I’m practicing:

Act as a small-business owner calling QBO support with a realistic bookkeeping problem. Don’t tell me the answer. Let me troubleshoot it by asking you questions.

That last one?

Trainer gold.

Now AI isn’t replacing the learner.

It’s giving the learner someone to practice on.

Step 3: Tell It How You Learn

This is where things start getting genuinely useful.

Not everybody learns the same way.

Maybe giant walls of technical text make your eyes cross.

Maybe you need examples.

Maybe you need repetition.

Maybe you understand something only after physically doing it.

Maybe you want somebody to explain the concept first and then let you go screw around in the sample company.

Tell the AI that.

For example:

I learn best by understanding why something works, seeing one realistic example, and then trying it myself. Don’t give me five examples at once. Teach me one concept, give me a practice task, and let me answer before continuing.

That’s not magic prompt engineering.

That’s just communicating.

Step 4: Give It Your Workspace

This is one of the biggest differences between occasionally using AI and actually making it useful.

Give it context about what you’re working on.

If your AI platform supports projects, workspaces, persistent instructions, uploaded reference material, or similar features, use them.

Your workspace might contain:

• Your current learning goal.
• Which ProAdvisor module you’re studying.
• Notes you’ve already taken.
• Terminology you’re learning.
• Training materials you’re allowed to use.
• Questions you got wrong.
• Practice scenarios.
• Your preferred teaching style.
• Things you already understand and don’t need explained again.

Now when you ask a question, you’re not starting from zero every single time.

Instead of:

What’s an unapplied payment?

You can ask:

I ran into unapplied payments again. Explain this using what we’ve already covered about Accounts Receivable.

That’s a completely different learning experience.

Step 5: Give It a Personality — Yes, Seriously

This part is optional.

It’s also the part people laugh at until they try it.

If you’re going to spend significant time learning with a tool, make the interaction tolerable.

You can tell it:

Be conversational and approachable. Humor is welcome. Don’t use corporate motivational language. If I’m wrong, tell me I’m wrong and explain why. Don’t praise every answer I give you. I want useful feedback, not a cheerleader.

That last part matters to me.

I don’t want:

Amazing job! You’re absolutely crushing it! 🌟

when I have just confidently misunderstood Accounts Payable.

Tell me I screwed it up.

Then teach me why.

My own AI assistant has developed a particular personality because I’ve used it extensively across wildly different parts of my life and work.

It knows I prefer normal-human explanations.

It knows I enjoy humor.

It knows I don’t want everything sanitized into corporate beige.

And that makes me more willing to interact with the material.

That’s not insignificant.

If learning is enjoyable, I’m going to do more of it.

Step 6: Make It Quiz You

This is one of my favorite educational uses of AI.

Don’t ask it for answers.

Make it withhold them.

Try:

Give me five beginner QBO navigation scenarios one at a time. Ask where I would go or what I would do. Do not reveal the answer until I respond. If I’m wrong, explain the reasoning and then give me a similar question before moving on.

Now you’ve created yourself a tutor.

Or try:

Give me a fictional small-business client with a bookkeeping problem related to the material I’m studying. I have to diagnose the issue by asking questions. Only give me information I specifically ask for.

Now you’ve created yourself a customer simulation.

This is where I think AI has enormous potential in training.

Not doing the thinking for people.

Giving people more opportunities to think.

Step 7: Make It Challenge You

AI has an unfortunate tendency to be agreeable.

That’s pleasant when you’re discussing dinner.

It’s considerably less useful when you’re trying to learn something.

So tell it not to.

Add instructions like:

Don’t assume my conclusion is correct just because I sound confident.

Or:

If I propose a solution, identify anything I may have overlooked before agreeing with me.

Or:

When I answer a question incorrectly, don’t immediately give me the answer. Give me one hint and let me try again.

Now we’re getting somewhere.

I don’t need AI to tell me I’m brilliant.

I need it to help me become better.

Step 8: Verify the Important Shit

This deserves its own section in enormous flashing letters.

AI OUTPUT IS NOT SOURCE DOCUMENTATION.

If you’re learning QBO and the AI tells you:

• where a current feature lives,
• how a current Intuit policy works,
• what the certification requires,
• how a tax rule applies,
• what an accounting standard requires,
• or anything else where being wrong actually matters,

verify it.

Use official Intuit documentation for current product behavior.

Use your official training materials for certification requirements.

Use appropriate authoritative sources for accounting and tax rules.

AI can help you understand the material.

It should not magically become the authority for the material.

That’s an important distinction.

Side Quest: Build Your Own AI Study Partner

Here’s your activity for this entry.

Open whichever AI tool you’re using and give it something like this:

I’m learning QuickBooks Online and working toward ProAdvisor certification. Act as my study and troubleshooting partner.

Explain accounting concepts in conversational language before connecting them to the QBO workflow.

I learn best through realistic examples and hands-on practice.

Ask me questions rather than automatically giving me answers.

If I’m wrong, explain what part of my reasoning is wrong and let me try again.

Don’t agree with me simply because I sound confident.

When information depends on current QBO features, certification requirements, accounting standards, or tax rules, remind me to verify it against an authoritative source.

Keep explanations focused unless I ask for more detail.

Humor is welcome. Corporate motivational speeches are not.

Then customize it.

Make it yours.

Maybe you want concise answers.

Maybe you want detailed ones.

Maybe you want flashcards.

Maybe you want quizzes.

Maybe you want increasingly difficult fictional clients.

Maybe you want it to explain everything as though you’re five.

Maybe you want it to swear when you finally understand reconciliation.

It’s your assistant.

Train the damn thing accordingly.

The Point of This Side Quest

I called this category Accounting Against the Machine, which makes writing an article about how useful AI can be wonderfully ironic.

But that’s also the entire philosophy.

I’m not interested in pretending AI doesn’t exist.

I’m not interested in banning it from learning.

I’m also not interested in letting it turn capable people into passengers.

Use AI to explain.

Use it to practice.

Use it to organize.

Use it to brainstorm.

Use it to create simulations.

Use it to quiz yourself.

Use it to discover what you don’t understand yet.

Then go into QBO and do the work yourself.

Learn what happened.

Learn why it happened.

Learn how to fix it when it goes wrong.

Because the goal isn’t to create a person who can ask AI how to solve every problem.

The goal is to create a person who understands the problem well enough to recognize when the AI is full of shit.

Use the machine.

Train the human.

And then get back to your ProAdvisor modules, Resistance Accountant.

We have shit to learn.