You’re Not Behind on AI. You’re Paying the Wrong Tax.

Here’s the math nobody runs on themselves.

Say you’re making $60,000 a year. That works out to roughly $30 an hour. But here’s the thing, you probably don’t have very many $30-an-hour tasks in your actual day. What you have is a handful of $100-an-hour tasks and a whole lot of $9-an-hour tasks, and somewhere along the line they average out to that $30.

The $100-an-hour tasks are the ones where your real value lives. The strategy. The decisions. The creative thinking that only you can do with the context and judgment you’ve built over years. Those are the moments you went into your field for.

The $9-an-hour tasks? Those are the data entry, the inbox management, the report building, the scheduling, the copy-pasting between systems. The stuff that has to get done, but doesn’t require you specifically to do it.

The problem isn’t that the $9-an-hour tasks exist. It’s that they’re eating your $100-an-hour time. And right now, that’s a solvable problem.

If you could hand off most of those $9-an-hour tasks and reclaim that time for the work only you can do, what would change about your job? What decisions would get better? What would you finally have time to actually build?

That’s what this is about. Not AI hype. Not replacing your job. Just fixing the math.

The hidden cost nobody’s calculating

Back in the early 2000s, I was managing a 200-room hotel property. I loved the strategic side of that job, building corporate accounts, driving occupancy, figuring out how to move the property. That’s where my value was.

But I was also placing inventory orders twice a week. That’s going through a list, clicking into a system, finding the SKU for each line item, filling out the form, repeat. For a property that size, that’s about 45 minutes per order. Twice a week. An hour and a half, just on that one task.

And that was one of maybe ten tasks like it.

By the time I added up all the data-entry-style work, the logging, the reporting, the inbox, the administrative stuff that had to happen, I was easily 15 hours into the week before I got to anything that actually required my brain.

That meant the decisions that needed strategy got made from the hip. Fast, without the research I should have done. Without the time to think through the options properly. The quality of those decisions, the ones that actually moved the property, suffered directly because the $9-an-hour work ate the time I needed for the $100-an-hour thinking.

That’s the real cost. Not just hours. The quality of your best work goes down because you’re running on whatever’s left after the grunt work is done.

The two frameworks that fix the math

When I think about handing off work to AI, I use two frameworks together. Not one, two. They do different things.

Framework one: the $9/$100 split. Go through what fills your actual day and sort it honestly. What’s $100-an-hour work? That’s the strategy, the decisions, the judgment calls, the creative thinking, the relationship-building, the things where your specific experience and perspective change the outcome. What’s $9-an-hour work? That’s anything where the task just needs to get done correctly, it doesn’t need you to do it.

The goal is to get AI handling as much of that $9-an-hour column as possible so you’re spending more time in the $100-an-hour column.

Framework two: the four-quadrant matrix. This one is a grid. One axis is like versus don’t like. The other axis is good at versus not good at. Everything you do falls somewhere in those four quadrants.

• Things you don’t like and aren’t good at: AI can have those completely. Stop doing them yourself.
• Things you don’t like but are good at: teach AI to do them the way you’d do them. Get yourself out of that column.
• Things you like but aren’t good at yet: use AI to help you get better faster. It can accelerate your learning there.
• Things you like and are good at: protect that time. That’s where you should be spending more of your day, not less.

Most people, when they hear “use AI,” immediately imagine the $100-an-hour work, the strategy, the creative thinking, the judgment calls. They picture handing the best parts of their job to a machine. That’s not what this is.

This is about identifying everything that’s crowding out the best parts of your job and systematically moving it off your plate.

What that actually looks like in practice

Here’s a real example. I run social media for a company doing over $700 million in revenue, with more than 100 locations. Part of that job is responding to comments across all of that content.

As a writer who understands the audience, I’m good at it. But it’s repetitive. You’re answering the same questions, responding to the same types of comments, over and over. And at the scale I’m working at, doing that manually was easily two hours a day.

So I built an AI agent to handle it. I gave it the best examples of responses to our most common comment types. I gave it detailed persona briefs for each of our audience segments. I connected it to our social media data.

Now the agent pulls a daily report. It reviews what’s come in, prioritizes by importance, and drafts replies in our brand voice. I spend about ten minutes reading the report, reviewing the drafts, tweaking anything that needs it, and posting.

Ten hours a week, compressed into thirty to fifty minutes.

And here’s the part that doesn’t show up in the time math: there’s a specific kind of cognitive weight that comes with knowing a comment section is waiting for you. The dread of it. The mental energy you spend just anticipating it before you even open the app. That’s gone too. I didn’t realize how much that was costing me until it wasn’t there anymore.

That’s what fixing the math actually feels like.

What AI actually is right now

Most people still think of AI as a smarter Google. You type something in, it gives you an answer, hopefully without making things up.

That’s not what it is anymore.

What you actually have access to right now is a team of digital employees. You can give them access to the tools you already use, your email, your project management system, your calendar, your social platforms. You can build a repository of context about who you are, how you work, who your clients or customers are, and how you like things done. You can give them specific skills that replicate the exact workflows you do yourself.

And then they do the work. Not give you information about the work. Not draft an outline for you to execute. Actually do it, and hand you a checkpoint to review.

That’s the shift most people haven’t made yet. They’re using AI as a search tool when they could be using it as a workforce.

Here’s where to start. Not someday. This week.

Step one: Run the two-framework audit on your own job.

Take 20 minutes. Write down everything that fills your actual day, not your job description, your actual day. Then sort each item: Is this $100-an-hour work or $9-an-hour work? Is this something I like or don’t like? Is this something I’m good at or not good at?

You don’t need perfect answers. You need honest ones. The patterns will show up fast.

Step two: Take your audit to Claude.

Go to Claude.ai, the free version works fine for this. Tell it who you are and what you do. Paste in your list. Tell it which things you like and don’t like, which things you’re good at and not good at.

Then ask it this specifically: “How could AI agents, using Claude’s desktop app or agentic tools, help remove me from 90% of my $9-an-hour tasks and free up more time for my $100-an-hour work?”

Before you hit send: tell it to ask you as many questions as it needs to until it’s 99% confident it can actually help. Then let it ask.

What it comes back with might surprise you.

Step three: Pick one thing and actually hand it off.

Don’t try to automate your entire job in a week. Pick the one task from your audit that you hate most and are most good at, the thing that takes real skill but brings you no joy, and spend an hour setting up an AI workflow to handle it.

That one task is your proof of concept. Once you feel what it’s like to get that time back, the next one is a lot easier to identify.

The math doesn’t fix itself. But once you start running it, it compounds fast.