The Part of Your Job AI Can’t Touch.

There's a difference between automating a task and automating an entire job.

And the elevator operator and airline pilot are two examples of how differently that can play out.

Automation made one role unnecessary. The other became part of a much larger industry, even as technology took over more of the work.

That distinction matters as AI becomes capable of doing more of YOUR work.

Because the question isn't simply what you can automate.

It's whether automation is making your role MORE valuable or LESS valuable, and what you can do about it.

The elevator operator and the airline pilot

In a 2016 study, economist James Bessen examined 270 occupations listed in the 1950 US Census.

Of those 270 jobs, only ONE had disappeared primarily because of automation by 2010.

The elevator operator.

For decades, elevators required someone to operate a lever that controlled their movement between floors.

Then automatic elevators came along.

Passengers could select their own floors, and computerized systems could coordinate elevators throughout entire buildings.

There wasn't much of a job left for the operator.

Now compare that to airline pilots.

Autopilot has been around since the early 1900s, and today's aircraft can handle much of the routine flying that pilots once did manually

These days, autopilot handles roughly 90% of the flying.

You'd think automating that much of the work would make pilots less necessary.

But the opposite happened.

The job didn't disappear. The number of pilots SKYROCKETED as commercial aviation expanded.

Why?

Because the pilot's job was never ONLY operating the controls.

It's getting passengers to their destination SAFELY.

That means monitoring the aircraft, making decisions, and stepping in when something goes wrong. A bird strike. Equipment failure. An emergency landing.

As aviation technology improved and air travel expanded, pilots remained essential to delivering that outcome.

The elevator operator's job was operating the mechanism. The pilot's job is owning the outcome.

Now think about your own work.

As AI takes over more of your daily tasks, are you becoming more valuable because you own the outcome? Or is it taking over the very work your role exists to perform?

The mistake is treating automation like an on/off switch

When people start using Claude, a common question is:

"Can I automate this?"

But that's only part of the question.

Take something like creating a report.

You collect information, organize it, analyze it, identify what's important, and turn your findings into something another person can use.

Claude can help with every step.

But extracting numbers from documents is very different from deciding what those numbers actually mean.

One has a relatively straightforward definition of success.

Did Claude pull the correct information?

The other requires judgment.

Are the numbers meaningful?

Does the conclusion make sense?

Did the analysis miss something important?

You can use AI for both, but they require different levels of supervision.

If you can clearly define what a correct result looks like, you can automate more of the execution.

And when Claude handles that work reliably, you can spend more time on the parts that make the report useful. Investigating unexpected results, identifying problems, or recommending what should happen next.

Now you're not simply producing the same report faster.

You're capable of delivering a BETTER one.

The Elevator Test: Where does your judgment belong?

So how do you apply this to your own work?

I'd start by writing down the steps involved in something you do regularly.

Then ask yourself three questions.

1. What can I automate?

Look for tasks where you can clearly define what a correct result looks like.

Extracting information, converting files, checking calculations, or organizing data according to specific rules.

These are good starting points for Claude because you can test whether the output is right.

Say you're using Claude to process invoices. Instead of asking it to extract the information, tell it:

Extract the vendor, date, invoice number, and amount from each invoice. Flag missing information rather than guessing, and identify any entries that need my review.

Then test the results against invoices you've already processed.

Once Claude consistently gets them right, you can spend less time entering information and more time identifying unusual expenses or catching billing problems.

2. Where do I need to stay involved?

Some work requires more than producing the correct information.

Claude can research competitors, analyze data, or recommend a course of action.

But someone still needs to decide whether the recommendation makes sense.

Does it account for the right information?

Are there risks it missed?

Would you actually stand behind that decision?

You can give Claude better instructions, provide examples, and require sources to improve the results.

But automating the execution doesn't mean automating your judgment.

3. What can I do MORE of?

If Claude handles more of the repetitive work, what can you take on that you couldn't before?

Maybe you can investigate more questions, catch problems earlier, or make recommendations instead of simply delivering information.

That's how AI starts expanding what you're capable of delivering.

What happens when you get this right?

The biggest opportunity with AI isn't automating every task until there's nothing left for you to do.

It's using automation to take on work you couldn't before.

When Claude handles more of the repetitive execution, you have more capacity to solve problems, make decisions, and improve what you deliver.

You don't need to remove yourself from the process to get more out of AI.

You need to stop spending so much time operating the controls and start focusing on where you can make the biggest difference.

So the next time you sit down to automate something with Claude, ask yourself:

What can I hand over, how will I know it's right, and what can I do MORE of as a result?

Cheers, see y’all in the next newsletter.

Whenever you're ready, here's how I can help:

For Business Owners & Executives:
1. Apply for my Executive AI Coaching Program: Linked here
 2. Build a Team of AI Agents that save you 10+ hours a week: Linked here

For Everyone:
3. Use BuildPartner.ai to build faster with Claude Code (try free): Linked here